home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!stanford.edu!rutgers!princeton!phoenix.Princeton.EDU!amlogan
- From: amlogan@phoenix.Princeton.EDU (Adam M. Logan)
- Newsgroups: sci.math
- Subject: Re: High Order Undulation Seeds and Binary Undulants
- Message-ID: <1992Jul24.171121.17030@Princeton.EDU>
- Date: 24 Jul 92 17:11:21 GMT
- References: <9207241316.AA14033@ucbvax.Berkeley.EDU>
- Sender: news@Princeton.EDU (USENET News System)
- Organization: Princeton University
- Lines: 32
- Originator: news@ernie.Princeton.EDU
- Nntp-Posting-Host: phoenix.princeton.edu
-
- In article <9207241316.AA14033@ucbvax.Berkeley.EDU> cliff@watson.ibm.com ("Cliff Pickover") writes:
- >"Binary undulants" are numbers of the form 2**N that undulate
- >(alternate) the digits 1 and 0 in their decimal expansion.
- >
- >For example, 2**[9]49 is the "highest quality" binary undulant so far
- >discovered because it has the undulating sequence 101010 in it:
- >
- >2**949 = [...?]
- >4758454107128905800953799994079681792420032645310062268978469949811010102913
- [...]>Here N = 949 is called an undulation seed of order 6, since it
- >gives rise to a 6-digit undulation pattern of adjacent 1's and 0's.
- >
- >Are undulation seeds of higher order easily found using either
- >mathematical prowess or computation power?
-
- Interesting problem...I'll try the "mathematical prowess". In fact, I'll
- restrict myself to finding powers of 2 that start with adjacent,
- alternating 1's and 0's. Let a_n be the n-digit number 10101..., and let
- b_n = log_10 (a_n), c_n = log_10 (a_n + 1). Of course, a necessary and
- sufficient condition that 2^k start with a_n is that the fractional part
- of log_10 (2^k) be at least that of b_n and less than that of c_n. But
- log_10 (2^k) = k log_10 (2), and it is known (since log_10 (2) is
- irrational) that this sequence is uniformly distributed mod 1, asymptotic-
- ally the proportion of powers starting in a_n is c_n - b_n, which, in fact,
- for decent-size a_n (say >= 10) is about 1/(a_n ln 10). Of course,
- similar results hold for other bases, and thjis works for any
- old digit string, not just a_n. A more sophisticated analysis
- could look for patterns within the number, of course, but this is good
- enough to show existence. As for finding them, compute the logs to high
- enough precision, etc.
-
- Adam
-