[seqfan] Re: automated discovery of unknown formulae

Marc LeBrun mlb at well.com
Tue Oct 25 17:46:54 CEST 2011


>="Max Alekseyev" <maxale at gmail.com>
> Run Superseeker on a dedicated server (with a local copy of the OEIS)
> on every sequence Axxxxxx and report matches with a sequence Ayyyyyy

Max, great project!

This is an expanded variation of something I've been advocating for years
(heh, including just yesterday, in the thread '"Suggested Projects" page on
the wiki').

I sometimes refer to the sets of sequences generated in this way as a
"superseeker neighborhood".

You seem to be describing computing the superseeker neighborhood of the
entire OEIS--nice and ambitious!

As an initial proof-of-concept, you could start smaller--say with just the
superseeker neighborhood of the "core" sequences.

I think if any of THOSE are missing, they should be added to the OEIS, and
all missing cross references between T_i(A_core) and A_core should be added.

(The entire core neighborhood being arguably "interesting" defers the issue
of having to filter the chaff by manual inspection.)

Also, besides these big one-time batch projects, I think checking the
neighborhood of every new submission should be spun off as a background
process and any discovered hits reported back.  In past days this would have
been prohibitive, but Moore's Law ultimately prevails.

> Is anybody interested in setting up such a project?

Actually doing it, aye, there's the rub!





More information about the SeqFan mailing list