home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!wupost!kuhub.cc.ukans.edu!husc-news.harvard.edu!zariski!kubo
- Newsgroups: sci.math
- Subject: Sieve (Was Re; Re; prime quadruplets ... partial apology)
- Message-ID: <1992Jul30.171603.14329@husc3.harvard.edu>
- From: kubo@zariski.harvard.edu (Tal Kubo)
- Date: 30 Jul 92 17:16:02 EDT
- References: <keRf_Ru00iUy02UVMt@andrew.cmu.edu> <1992Jul29.222749.18849@spool.cs.wisc.edu>
- Organization: Dept. of Math, Harvard Univ.
- Nntp-Posting-Host: zariski.harvard.edu
- Lines: 31
-
- In article <1992Jul29.222749.18849@spool.cs.wisc.edu>
- bach@jalapeno.cs.wisc.edu (Eric Bach) writes:
- >
- > Questions like this have received a fair amount of study. The
- > conjectured density of quadruplets <= N is ~ C * N / (log N)^4,
- > where C is expressible as an infinite product over the primes.
-
- 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?
-
- 2) Can one guess the prime number theorem from heuristics?
-
- > 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
- to believe that breakdown of heuristic hypotheses such as
- independence could be tested in the range of computations
- feasible today?
-
-
- tal kubo@zariski.harvard.edu
-
-