[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference rusure::math

Title:Mathematics at DEC
Moderator:RUSURE::EDP
Created:Mon Feb 03 1986
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2083
Total number of notes:14613

1350.0. "a problem with primes" by GUESS::DERAMO (Sometimes they leave skid marks.) Fri Dec 07 1990 10:53

Path: ryn.mro4.dec.com!shlump.nac.dec.com!decuac!haven!ames!elroy.jpl.nasa.gov!usc!zaphod.mps.ohio-state.edu!rpi!bu.edu!cvbnet!jersey.uucp!dwilson
From: [email protected] (David Wilson {x6203})
Newsgroups: sci.math
Subject: Prime problem
Message-ID: <[email protected]>
Date: 6 Dec 90 17:27:42 GMT
Sender: [email protected]
Reply-To: [email protected] (David Wilson {x6203})
Organization: Computervision Div., Prime Computer Inc., Bedford, MA
Lines: 32
 
 
 
    Let x be an integer >= 2.  Define s(x) to be the smallest prime
    factor of x, and let f(x) = x+s(x)-1.
 
	1.  Show that gcf(x, f(x)) = 1.
 
    Now, given integer x >= 2, define sequence a[x] by taking f
    repeatedly on x.  a[2] begins with the values
 
	2*, 3*, 5*, 9, 11*, 21, 23*, 45, 47*, 93, 95, 99, 101*, ...
 
    Note that a[2] is fairly heavily laced with primes.  However,
 
	2.  Show that every a[x] contains an infinite number of
	composite elements
 
    and, indeed
 
	3.  Show that for each N >= 0, there exists an x such that the
	first N elements of a[x] are all composite.
 
    Finally
 
	4.  Does every a[x] contain a prime number?  An infinite number
	of prime numbers?
 
 
David W. Wilson ([email protected])
J.H.Whitney (was Prime Computer (was Computervision Corp.)), Bedford, MA
Disclaimer: "Truth is just truth...You can't have opinions about truth."
- Peter Schikele, introduction to P.D.Q. Bach's oratorio "The Seasonings."
T.RTitleUserPersonal
Name
DateLines
1350.1HERON::BUCHANANHoldfast is the only dog, my duck.Sat Feb 09 1991 13:32109
Comment:   This is an interesting puzzle, where you have to look at the
function differently for each question posed.   The first half of Q4,
remains open.

>    Let x be an integer >= 2.  Define s(x) to be the smallest prime
>    factor of x, and let f(x) = x+s(x)-1.
> 
>	1.  Show that gcf(x, f(x)) = 1.

	Let g(x)
=	gcf(x, f(x))
=	gcf(x,x+s(x)-1)
=	gcf(x,s(x)-1)

	Suppose p(x) is the smallest prime factor of g(x).   Then p(x) is a 
prime factor of x, but also p(x) | s(x)-1, and hence p(x) < s(x).   This 
contradicts the minimality of s(x).   So no such p(x) exists, so g(x) = 1.

>    Now, given integer x >= 2, define sequence a[x] by taking f
>    repeatedly on x.  

	Remark: 1 cannot appear in any such a[x], nor can 2 in any position
apart from initially in a[2].

> a[2] begins with the values
>
>	2*, 3*, 5*, 9, 11*, 21, 23*, 45, 47*, 93, 95, 99, 101*, ...
> 
>    Note that a[2] is fairly heavily laced with primes.  However,
> 
>	2.  Show that every a[x] contains an infinite number of
>	composite elements

	It is sufficient to show that given any x, the sequence a[x]
contains a composite element.   Assume the contrary.   Then there is a prime y 
such that the sequence:
	y, f(y), f(f(y)), f(f(f(y))), ...
consists entirely of primes.   y >= 2, by the remark above.   For p prime 
s(p) = p, and hence f(p) = 2p-1.   Thus writing z = y-1, the series is:
	z+1, 2z+1, 4z+1, 8z+1, ...
Now, y | 2^z - 1, for all prime y > 2, by a well-known theorem, and hence:

	y | z*(2^zk-1) + z+1	for all k >= 1
=>	y | (2^zk)z + 1		for all k >= 1

	Since a[y] is monotonically increasing, this means that every zth
term, after the initial, is composite.   In particular the zth term is
composite, proving a contradiction.   

	Thus every sequence a[x] contains a composite, and hence an infinite 
number of them.


>    and, indeed
>
>	3.  Show that for each N >= 0, there exists an x such that the
>	first N elements of a[x] are all composite.

	Let a[x](j) denote the jth term of a[x].   In particular a[x](0) = x.
Let m = a[2](N-1).   Then the first N terms of a[m!+2] will be composite.

Proof:
	I claim that a[m!+2](j) = m! + a[2](j), for j=0,...,N-1.
This is true by induction.   For j=0, it is trivial, and so now assume that
we know it to be true of j-1.
	a[m!+2](j) = f(m!+a[2](j-1)).   s = s(a[2](j-1)) is the smallest prime
to divide a[2](j-1).   s | m!, since s =< a[2](j-1) < m (by monotonicity of 
the sequence a[.]).   But for any prime p < s, p | m! for the same reason, and
so (mod p) m! + a[2](j-1) == a[2](j-1) <> 0, by definition of s.   So s is
also the smallest prime to divide m!+a[2](j-1).   Therefore f(m!+a[2](j-1))
= m! + a[2](j-1) + s - 1 = m! + a[2](j), as required.

	From this we observe trivially that for j=0,...,N-1:
(i)	a[2](j) < a[m!+2](j) and
(ii)	a[2](j) | a[m!+2](j)

which together are enough to show that a[m!+2] is composite for the first N
terms.

 
>    Finally
> 
>	4.  Does every a[x] contain a prime number?  An infinite number
>	of prime numbers?

	Remark:   Once again, if every a[x] contains a prime number, it must
contain an infinite number of prime numbers.   So the answer to the second 
question is the same as the answer to the first.

	Let s[x] be a sequence derived from a sequence a[x], and defined as 
s[x](j) = s(a[x](j)).   Is it possible that s is periodic, and hence we can
avoid a[x] containing a prime in this way?   The answer is no.

	Suppose that s[x] only contains a finite number of different values, in
the set V = {v(i),i=1,...,I}.   Then consider v = product(i) v(i).   If 
a[x](j) < kv, for some constant k, then there is no way that a[x](j') >= kv, 
for any j' > j.   To see this observe that if s(x) = s, then x =< kv - s (the 
largest multiple of s less than kv) and hence f(x) =< kv - 1.   But a[x] is
monotonically increasing to infinity (since a[x](j+1) >= a[x](j)+2), so we have
a contradiction.   Hence, s[x] contains an infinite number of different values.

	Let's extend this idea.   Let p be any prime, and let p? denote the
product of all odd primes up to and including p.   Any time that
f(x) div p? > x div p?, we must have that s(x) > p.   This isn't sharp enough.
Any ideas?   My guess is that every sequence must spit out a prime eventually,
but I can't show it.

Regards,
Andrew.