[seqfan] Re: about 10000 numbers, a certain document from M.I. Shamos.

[Richard Mathar]

On my wish list --as a physicist with a desire to find values of integrals,
sums and other non-integer stuff-- is a database like OEIS (that is, maintained,
with an option to submit new values) which does this for the
rationals/constants/irrationals.
As Shamos correctly points out, the Plouffe database does not match the
interest that stems from some real-world applications, but is a rather
artificial set. The mix of values like Madelung constants, zeta-values,
special functions (Gamma, Hypergeometric, Elliptic integrals) is statistically
different from what is offered by these numbers.
   In the past, I have submitted these values to the OEIS. That works
well, and the "keyword:cons" is essentially all what is desired,
but the fact that the OEIS can "revert" floating point numbers
as good as integers is not that widely known. Note that the question
of a "finite precision" of these does certainly *not* arise in practise.
  In addition, nobody of the audience that I have in mind would search
for a integer representation (continued fraction, Engel expansion,...)
in favor of the decimal representation. The idea do attack these series
of digits with Mobius transforms is ... blasphemy..

Richard