home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!stanford.edu!CSD-NewsHost.Stanford.EDU!Sunburn.Stanford.EDU!pratt
- From: pratt@Sunburn.Stanford.EDU (Vaughan R. Pratt)
- Subject: Re: exp(pi*sqrt(x))
- Message-ID: <1992Nov23.001203.20604@CSD-NewsHost.Stanford.EDU>
- Sender: news@CSD-NewsHost.Stanford.EDU
- Organization: Computer Science Department, Stanford University.
- References: <1992Nov22.124131.17689@husc15.harvard.edu>
- Date: Mon, 23 Nov 1992 00:12:03 GMT
- Lines: 32
-
- In article <1992Nov22.124131.17689@husc15.harvard.edu> blom@husc15.harvard.edu writes:
- >Aitken once remarked that exp(pi*sqrt(163)) differs from an integer by
- >less than 10^-12. Why and when does exp(pi*sqrt(x)) approximate an integer?
-
- Q(sqrt(-163)) (the field of rationals extended to complex quadratic
- field with sqrt(-163)) is a unique factorization domain ("class number
- 1"). The 9 numbers that work here are 1 2 3 7 11 19 43 67 163, the
- larger ones work much better for getting good integer approximations
- this way.
-
- Though these 9 were part of a case of a conjecture due to Gauss, it was
- discovered only in 1933 that there were no others in the range -164 to
- -5e9. In 1934 it was shown that there was no 11th number---at most one
- remained to be discovered. By 1966 the -5e9 bound had been improved to
- -1e(9e6) (10 to the 9 million), i.e. if there was a tenth it was below
- that limit. In 1966 H. Stark and A. Baker independently ruled out the
- possibility of a tenth altogether. [Source: Harold Stark's "Intro to
- Number Theory", p.295]
-
-
- Problem: characterize those powers of exp(pi*sqrt(163)/3) that
- approximate an integer (to better than .1% say). Why is it about half
- of the first 24, and seemingly none of the rest (except for the
- necessary one in a thousand due to the .1% tolerance)?
-
- If you want a very convenient way to play around with these numbers
- interactively, and have a Sun-4, a binary of Hans Boehm's
- infinite-precision calculator can be found on boole.stanford.edu as
- /pub/calc. To get the base power, start it and type 163rp*3/x to get
- 640320.00000000060486373504901603947174181881853947577148576036659181946522182
- --
- Vaughan Pratt A fallacy is worth a thousand steps.
-