home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!darwin.sura.net!jvnc.net!princeton!lentil.princeton.edu!jwr
- From: jwr@lentil.princeton.edu (Jaroslaw Tomasz Wroblewski)
- Subject: Re: Primes in x_{n+1} = ax_n+b (was Re: u(v^n)w prime puzzle)
- Message-ID: <1992Aug21.180614.17635@Princeton.EDU>
- Sender: news@Princeton.EDU (USENET News System)
- Nntp-Posting-Host: lentil.princeton.edu
- Organization: Princeton University
- References: <a_rubin.714331997@dn66> <1992Aug21.103132.29967@ecrc.de>
- Date: Fri, 21 Aug 1992 18:06:14 GMT
- Lines: 43
-
- In article <1992Aug21.103132.29967@ecrc.de> jeanmarc@ecrc.de writes:
- >Let a,b be integers, and (x_n) be a sequence s.t. x_{n+1} = a x_n + b
- >
- >My question is: does it contain infinitely many primes ?
- >
- >There are trivial cases, where the answer is no:
- >1/ if (a x_0 + b = x_0) the sequence is constant.
- >2/ if (a = -1) the sequence alternates between two values x_0 and x_1.
- >3/ if ((a,b) > 1 or (x_0,b) > 1) then the sequence clearly contains only composite numbers.
- >
- >What about the other cases ?
-
- For any odd b we can see easily that x_n=2^n-b satisfies the relation
- x_{n+1} = a x_n + b with a=2 and is not a trivial case above.
-
- For b=1 we get Mersene numbers. Most people will bet there are infinitely
- many primes here, but I would not expect any proof coming.
-
- However there exist odd b such that the sequence x_n=2^n-b contains finitely
- many primes. That is the case because one can pick b in such a way that
- For n=1(mod 2) 3|2^n-b
- For n=2(mod 4) 5|2^n-b
- For n=0(mod 3) 7|2^n-b
- For n=4(mod 12) 13|2^n-b
- (So far everything except 8(mod 12) is covered)
- For n=0(mod 8) 17|2^n-b
- For n=20(mod 24) 241|2^n-b
-
- Then for every n the number 2^n-b will be divisible by at least one of the
- primes 3,5,7,13,17,241.
-
- Therefore 1 and the above determined b will give different answer to your
- question. This shows that even making a good guess when such sequences
- contain infinitely many primes is tricky.
-
- Once we pick b as above I do not see a reason why other sequences
- satisfying x_{n+1} = 2 x_n + b should also contain only finitely many
- primes. Knowing a and b alone may not be enough to decide how many primes
- does a sequence have.
-
- --
-
- Jarek (Jaroslaw Tomasz Wroblewski) , E-mail jwr@math.Princeton.EDU
-