OEIS "Top N" lists?
Leroy Quet
qq-quet at mindspring.com
Fri Dec 5 03:32:22 CET 2003
Or perhaps sequences could be judged upon the integers themselves.
It might be interesting to rank sequences by, say, the sequences with the
fewest number of terms in common with other sequences (compared among
terms appearing in the database, of course).
..Or by the most uncommon absolute-differences among terms, the sequences
whose terms average the highest, the 'most chotic' sequences (ie. least
able to fit to a low-order polynomial, or some other definition of 'most
chaotic'), the sequences which contain the most terms of other sequences
among the terms listed (but in some non-respective order as to avoid
trivial cases), the sequences which contain the longest runs of
consecutive integers but which miss the occasional integer (ranked based
upon some formula involving the lengths of the integer-runs and the
number of ommitted integers among the consecutive integers), etc etc...
Perhaps rankings with more serious motives can be created, such as those
which inspire research into connections between sequences, connections
that would not have been obvious at first to the casual OEIS user without
the computer-aided analysis.
thanks,
Leroy Quet
>In looking at the comments on A002620 just now I'm reminded of why the OEIS
>is such an exciting repository of cultural knowledge. There are about a
>dozen very different interesting views on the sequence, contributed by as
>many authors. It's really something to be proud of.
>
>This got me to wondering if someone with the appropriate software might
>want to find the "Top N" (for N=5 or 10?) sequences in various categories:
> Longest comments
> Most references
> Most cited reference (ie bibliography entry)
> Most links
> Most contributors (by counting the (AT)s)
> Longest formulas (or most formulas? tricky to count!)
> Most "see alsos" to other sequences
> Most "see alsos" from other sequences
> Largest explicitly given integers (ie as 99999999 versus 10^10-1)
> etc
>(with possible sub-categories based on keywords, such as tabl etc)
>
>Perhaps along the way it could collect stats of various kinds (eg averages
>etc).
>
>It would be fun to rerun this census at different times, (eg 1/1/2004,
>A100000, etc), to see how these things might change.
More information about the SeqFan
mailing list