home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!crdgw1!ge-dab!puma.ATL.GE.COM!puma!sjameson
- From: sjameson@atl.ge.com (Stephen M Jameson)
- Newsgroups: comp.ai
- Subject: Re: Kolmogorov complexity
- Message-ID: <SJAMESON.92Aug13090132@fergie.atl.ge.com>
- Date: 13 Aug 92 13:01:32 GMT
- References: <1992Aug11.020517.18834@kronos.arc.nasa.gov>
- Sender: news@puma.ATL.GE.COM (USENET News System)
- Organization: General Electric Advanced Technology Labs
- Lines: 39
- In-Reply-To: wray@kronos.arc.nasa.gov's message of 11 Aug 92 02:05:17 GMT
-
- In article <1992Aug11.020517.18834@kronos.arc.nasa.gov> wray@kronos.arc.nasa.gov (Wray Buntine) writes:
-
- > > I hope someone will follow up with some indication of whether anyone
- > > has actually cared about the constant factors.
- >
- > The constant factor is all important because in practice we only have small
- > sample sizes. People who gloss over it usually have no experience at
- > competitive implementation (implementing an algorithm that *outperforms*
- > the *existing* competitors).
- >
-
- I attended undergraduate school at Johns Hopkins and graduate school at U of
- Maryland. The CS cirriculum at Hopkins was quite theoretical, while the basic
- graduate cirriculum at UofM was eminently practical, and I had a couple of
- lessons in putting asymptotic complexity into perspective. In my first
- graduate course at UofM, we were discussing various data structures/algorithms
- and reasons why one might prefer them and the professor indicated that option A
- was better because option B required 2 extra bytes per structure for an
- additional pointer or something like that. My initial reaction was "who cares,
- it's a constant factor," but the professor was not impressed.
-
- In the second course, we were discussing a one point an algorithm which
- required finding the mean of a set of numbers. He pointed out that there is a
- worst-case O(n) algorithm known for doing this, but "only an idiot would use
- it" because it is horrendous to implement, and the one he suggested was far
- simpler, and gave O(n) performance except in pathological cases. I gulped,
- having learned about the worst-case O(n) at Hopkins and having spent over a
- week implementing and debugging it for a program I wrote on a summer job.
-
- --
- Steve Jameson General Electric Aerospace
- sjameson@atl.ge.com Advanced Technology Laboratories
- Moorestown, New Jersey
- ****************************************************************************
- ** . . . but I do not love the sword for its sharpness, nor the arrow **
- ** for its swiftness, nor the warrior for his glory. I love only that **
- ** which they defend . . . **
- ** -- Faramir, "The Two Towers" **
- ****************************************************************************
-