home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!howland.reston.ans.net!spool.mu.edu!agate!linus!linus.mitre.org!gauss!bs
- From: bs@gauss.mitre.org (Robert D. Silverman)
- Subject: Re: Number Theory
- Message-ID: <1993Jan26.132347.4704@linus.mitre.org>
- Sender: news@linus.mitre.org (News Service)
- Nntp-Posting-Host: gauss.mitre.org
- Organization: Research Computer Facility, MITRE Corporation, Bedford, MA
- References: <1993Jan26.021132.15583@murdoch.acc.Virginia.EDU>
- Date: Tue, 26 Jan 1993 13:23:47 GMT
- Lines: 49
-
- In article <1993Jan26.021132.15583@murdoch.acc.Virginia.EDU> pjp9q@honi2.acc.Virginia.EDU (Paul Joseph Perrone) writes:
- :
- :I am the non-mathematician who posted the question regarding prime
- :numbers a few messages ago, and I apologize for incorrectly phrasing
- :my question.
- :
- :Here is my rephrased question:
- :
- :What is the LARGEST set of prime numbers fulfilling the following
- :criterion,
- : 1) No prime number in this set can be greater than 1 million;
- : 2) No sum nor difference between any two of the primes in this
- : set can be equal to the sum or difference between any other
- : combinations of primes in this set.
-
- This is a very interesting and unsolved problem. Firstly, let me say
- that it is sufficient to consider differences only. If p+q = r+s, then
- p-r=s-q.
-
- There is a heuristic which suggests that if M is the upper bound, then
- the size of the largest set is between O(sqrt(M)/log(M)) and O(sqrt(M)),
- but this is a heuristic only. Hint: assume primes are uniformly distributed
- odd numbers between 3 and M and look at the probability that 4 random
- variables drawn from this distribution satisfy x1+x2 = x3+x4. This
- upper bound is also suggested by the birthday paradox. One needs
- O(sqrt(M)) numbers approximately to give a 50% (roughly) chance of collision.
- There are other approaches based upon estimates from Goldbach's conjecture.
- These involve estimates of the number of ways 2N can be written as the sum of
- two primes for large N. These estimates also give an upper bound of O(sqrt(M))
-
- Thus the answer to your question is "around 1000". The exact number would be
- very difficult to calculate.
-
- :I am looking for references or direction in how to compute such a
- :maximum set for the above problem, and for cases where I might
-
- The maximum set for M might be very different for (say) M+2 if M+2
- is prime, for example. I do not know offhand how to find the set other
- than by exhaustive search, but can suggest a starting point:
-
- Consider the set 2,3,5,11,23,47 .... [double then add 1]. Consider the
- nearest prime to each term. This gives a starting set of O(log M) numbers.
- Then start "fitting" additional primes.
-
- --
- Bob Silverman
- These are my opinions and not MITRE's.
- Mitre Corporation, Bedford, MA 01730
- "You can lead a horse's ass to knowledge, but you can't make him think"
-