home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!stanford.edu!agate!agate!hughes
- From: hughes@soda.berkeley.edu (Eric Hughes)
- Newsgroups: sci.crypt
- Subject: Random vs. exhaustive search
- Date: 13 Oct 92 13:59:54
- Organization: /accounts/hughes/.organization
- Lines: 16
- Message-ID: <HUGHES.92Oct13135954@soda.berkeley.edu>
- References: <1992Oct13.115032.1@kean.ucs.mun.ca>
- NNTP-Posting-Host: soda.berkeley.edu
- In-reply-to: jgarland@kean.ucs.mun.ca's message of Tue, 13 Oct 1992 14:20:32 GMT
-
- In article <1992Oct13.115032.1@kean.ucs.mun.ca>
- jgarland@kean.ucs.mun.ca writes:
-
- >Also to search an entire keyspace when sampling _with_
- >replacement--which is the case in at least the one implementation of
- >genetic algorithms that I've seen--would be inordinately
- >time-consuming.
-
- Incorrect. The expected time of a random search is N. The expected
- time of an exhaustive search is N/2. There is no upper bound on the
- time of a random search. The upper bound on exhaustive search is N.
-
- Hence it really only takes twice as long. See the article on the
- Chinese Lottery I mentioned earlier. It is exactly this argument.
-
- Eric
-