home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!psinntp!kepler1!andrew
- From: andrew@rentec.com (Andrew Mullhaupt)
- Newsgroups: sci.math
- Subject: Re: Looking for help
- Message-ID: <1455@kepler1.rentec.com>
- Date: 9 Jan 93 18:25:59 GMT
- References: <1993Jan4.140128.20155@choate.edu> <1444@kepler1.rentec.com> <1993Jan8.150941.12619@infodev.cam.ac.uk>
- Organization: Renaissance Technologies Corp., Setauket, NY.
- Lines: 48
-
- In article <1993Jan8.150941.12619@infodev.cam.ac.uk> cet1@cus.cam.ac.uk (C.E. Thompson) writes:
- >In article <1444@kepler1.rentec.com>, andrew@rentec.com (Andrew Mullhaupt) writes:
- >|> In article <1993Jan4.140128.20155@choate.edu> jburne@spock.uucp (John Burnette)
- >|> writes:
- >|> >"Give two numbers, x and y, whose sum is rational and whose product is
- >|> >irrational."
-
- >|> >x = .10100100010000....
- >|> >y = .01011011101111....
-
- >|> >Obviously x+y=1/9 (and, please, let's not start that thread again...)
- >|> >but I've always been at a lost about x*y.
-
- >|> It isn't entirely clear but I suppose that x is a Liouville* number, in
- >|> which case it is transcendental.
-
- >It wasn't explicitly stated, but I took the chains of zeroes to be increasing
- >linearly in size, i.e. x = \sum_{n=1}^{\infty} 10^{-n(n+1)/2}. In which case
- >the gaps aren't big enough to make the rational approximations with denominator
- >10^{n(n+1)/2} convergents to x, let alone make it a Liouville number.
-
- >|> *Or a Thue-Siegel-Roth number, etc.
-
- >Same applies.
-
- Well if you want to talk about _that_ number life is indeed much more
- difficult. As I said, it isn't clear which number was originally given,
- so I gave the student the benefit of the doubt. But things can be done
- with the number you think x is. As noted in my previous post, the
- only way for x * (1/9 - x) to be rational with x irrational is for x to
- be a quadratic irrationality. By Euler-Lagrange, the continued fraction
- of such x will be periodic. If we put x = sum 10 ^ ((-n^2-n)/2) as you
- suggest, we get the continued fraction:
-
- x = [0,9,1,9,11,9,11,90,1,9,1,90,110,9,12,4,1,3,1,4,2,1,4,4,10,
- 1,10,9,907,2,1,2,6,3,1,4,1,9...]
-
- which does not appear to have a periodicity. We should not expect the
- continued fraction of x to be _purely_ periodic, since x < 1, but
- eventually periodic.
-
- Later,
- Andrew Mullhaupt
-
- P.S. I think you can get a handle on your number x using the theta
- functions, and then known transcendence results (Cf. Baker's
- _Transcendental Number Theory_) may apply, but I don't have time to chase
- this down.
-