home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!morrow.stanford.edu!leland.Stanford.EDU!ilan
- From: ilan@leland.Stanford.EDU (ilan vardi)
- Subject: Re: FAREY
- Message-ID: <1992Oct16.211940.28546@leland.Stanford.EDU>
- Sender: news@leland.Stanford.EDU (Mr News)
- Organization: DSG, Stanford University, CA 94305, USA
- References: <98687@bu.edu> <168827A63.DUENTSCH@dosuni1.rz.uni-osnabrueck.de> <1992Oct16.170803.14345@nntp.uoregon.edu>
- Date: Fri, 16 Oct 92 21:19:40 GMT
- Lines: 40
-
- In article <1992Oct16.170803.14345@nntp.uoregon.edu> scavo@cie.uoregon.edu (Tom Scavo) writes:
- >In article <168827A63.DUENTSCH@dosuni1.rz.uni-osnabrueck.de> DUENTSCH@dosuni1.rz.uni-osnabrueck.de writes:
- >>In article <98687@bu.edu> rehman@math.bu.edu (Naved Rehman) writes:
-
- >Ian Stewart also mentions the Farey sequence in his _Does God Play Dice?_
- >(Basil-Blackwell, 1989?).
-
- On the other hand, I would not trust Ian Stewart about such things.
- In one of his books he states that the Riemann hypothesis is
- equivalent to
-
- pi(x) = x/log(x) + O(sqrt(x) log^2(x))
-
- so he is clearly writing about something that he has no clue about.
- In fact, de la Valle Poussin showed that
-
- pi(x) = li(x) + O(x exp(-c sqrt(log(x))))
-
- where
-
- li(x) = int_2^x 1/log(x) dx
-
- and integration by parts gives
-
- li(x) = x/log(x) + x/log^2(x) + ...
-
- + (k-1)! x/log^k(x) + O(x/log^{k+1}(x))
-
- so
-
- pi(x) = x/log(x) + x/log^2(x) + O(x/log^3 (x)).
-
- The point is that a popular account of something as well understood
- and easy to look up as this should be extremely accurate.
-
- Moreover, I find the way he writes his column in Scientific American
- to be totally insulting. If the material were at all interesting then
- there would be reason to dress it up as some fable.
-
- -ilan
-