home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!sun-barr!ames!agate!linus!linus.mitre.org!faron!bs
- From: bs@faron.mitre.org (Robert D. Silverman)
- Subject: Re: Sieve (Was Re; Re; prime quadruplets ... partial apology)
- Message-ID: <1992Jul30.234102.1645@linus.mitre.org>
- Sender: news@linus.mitre.org (News Service)
- Nntp-Posting-Host: faron.mitre.org
- Organization: Research Computer Facility, MITRE Corporation, Bedford, MA
- References: <keRf_Ru00iUy02UVMt@andrew.cmu.edu> <1992Jul29.222749.18849@spool.cs.wisc.edu> <1992Jul30.171603.14329@husc3.harvard.edu>
- Date: Thu, 30 Jul 1992 23:41:02 GMT
- Lines: 46
-
- In article <1992Jul30.171603.14329@husc3.harvard.edu> kubo@zariski.harvard.edu (Tal Kubo) writes:
-
- stuff deleted....
-
- :i.e. the conjectured density is the product of local densities
- :[essentially (1-4/p)] taken over few enough primes less than N
- :to maintain the 'independence' of congruences modulo different
- :primes. Some questions for the experts:
- :
- :> This is based on probabilistic arguments,
- : ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
- : 1) Are there problems where the heuristic arguments fail?
- :
- Yes. Certainly. One very well known failure is the probabalistic argument
- for Fermat's Last Theorem. While it works for exponents greater than 3,
- it fails for exponent 3. That is to say the heuristic predicts that
- there should be an infinite number of solutions for exponent 3.
-
- : 2) Can one guess the prime number theorem from heuristics?
-
- Yes, but only approximately (up to a constant factor). See, for example
- the derivation of Merten's Theorem in Hardy and Wright. Roughly speaking
- the probability that a large number is prime is the probability that it
- is NOT divisible by 2,3,5,....... . This probability is (1-1/2)*(1-1/3)*
- (1-1/5)*.....(1-1/sqrt(n)). This product is O(n/log(n)). [the constants
- are known. see Merten's theorem]. There are other heuristics that suggest
- the constant is 1.
-
- :> and agrees with
- :> numerical data, as far as we know. Using sieve methods, it can
- :> be shown that the number is O( N / (log N)^4 ).
- :
- : 3) How convincing is the numerical data? Is there reason
-
- It is VERY convincing. In fact, in a loose sense it is overwhelming.
-
- : to believe that breakdown of heuristic hypotheses such as
- : independence could be tested in the range of computations
- : feasible today?
-
- Not really. If they do break down it is likely out of range.
- --
- 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"
-