[seqfan] Re: "Suggested Projects" page on the wiki

Charles Greathouse charles.greathouse at case.edu
Fri Oct 28 15:28:48 CEST 2011

> Also, memory has gotten much cheaper, but it still isn't free and unlimited.

Right.  Bandwidth is probably the scarcer resource: although we could
afford a hard drive array large enough to have a b-file for every
sequence (stopping at 10,000 terms, when no more terms are known, or
when the next term would have more than a thousand digits -- probably
a few TB) actually transferring more than a tiny sliver of a percent
of them would no doubt suck up all of our capacity.  Just think of how
much bigger b-files are than sequences!

Charles Greathouse
Case Western Reserve University

On Fri, Oct 28, 2011 at 1:58 AM,  <franktaw at netscape.net> wrote:
> I don't think we want to get too much into the OEIS computing things for us.
> That isn't its primary focus, and tools for which it is the primary focus
> are always going to be better at it. Beyond a certain point, you download
> the sequence into a tool like PARI or Mathematica, or even into a
> spreadsheet, and work with it there.
> Also, memory has gotten much cheaper, but it still isn't free and unlimited.
> Mathematics is one field of endeavor that can easily blow away any plausible
> computational limits.
> Franklin T. Adams-Watters
> -----Original Message-----
> From: Tom Duff <td at pixar.com>
> "T. D. Noe" <noe at sspectra.com> writes
>> As I said before, limits are a good thing when it comes
>> to b-files.  There are many reasons for this:
>> In general, every doubling of the b-file size cuts the
>> response time in half.  For many sequences, it is easy
>> to submit b-files with 10^6 terms.  Is seeing that many
>> terms good?  No, it is a waste of people's time to wait
>> for such a large file.  And you lose detail when you view
>> a file having more than about 10000 terms.
> There's no reason, other than sucking up more of Russ's valuable time,
> that oeis.org couldn't allow uploading of b-files of arbitrary
> length, but by default only deliver an initial subsequence, indicated,
> say by tweaking the b-file URL syntax.  For example A001235's
> b-file link is http://oeis.org/A001235/b001235.txt.  We could have
> http://oeis.org/A001235/b001235.txt?count=100 return just the first
> 100 terms, and present that URL as well wherever the b-file link is
> normally presented.  (And we could make compression optional, by something
> like http://oeis.org/A001235/b001235.txt?count=inf,compression=zip.  You
> can imagine tweaking this idea indefinitely. Formatting options? Arbitrary
> bases? Superseeker transformations?)
> --
> Tom Duff.  Probably does everything except what you want it to do.
> _______________________________________________
> Seqfan Mailing list - http://list.seqfan.eu/
> _______________________________________________
> Seqfan Mailing list - http://list.seqfan.eu/

More information about the SeqFan mailing list