OEIS Keywords

[franktaw at netscape.net]

I certainly agree that the current keyword system is a mess in some 
places.  Herewith my cut on the existing keywords:

base: dependent on base used for sequence

This definition is quite unclear, and the application is wildly 
variable.

bref: sequence is too short to do any analysis with

Clear enough.  This one is usually applied by Neil.

cofr: a continued fraction expansion of a number

Mostly clear.  There are some cases where this has been applied to
the numerators and/or denominators of the convergents.  I find 11
matches in the database for "keyword:cofr keyword:frac".

cons: a decimal expansion of a number

This also seems clear enough, although it need not be "decimal" - any 
base will do.
Also, a note that the "base" keyword should not also be added, unless 
bases were
used in defining the number itself.

core: an important sequence

This is a judgement call, but less problematic than other judgement 
calls.  Very rarely
does a new sequence get this keyword - if it was really fundamental, 
why wasn't it
there before.

dead: an erroneous sequence

Applied by Neil when errors are discovered.  Works fine.

dumb: an unimportant sequence

A judgement call, and one of the most difficult ones.  I don't think 
this adds any real
value to the database.

dupe: duplicate of another sequence

Currently used only for 1 sequence, A048789.  Not really very useful; 
let the comments
say it.

easy: it is very easy to produce terms of sequence

This is another one with wildly varying usage.  Sometimes this means: 
"the sequence
has a simple expression for the Maple combinatorics package"; trying to 
calculate it by
hand might take hours.  I don't think this one adds much value, either.

eigen: an eigensequence: a fixed sequence for some transformation - see 
the files transforms and transforms (2) for further information.

While there are some borderline cases, this is clear enough.  
Unfortunately, there are
many sequences it might apply to for which it is not present.  Might be 
useful if it was
uniformly applied.

fini: a finite sequence

Straightforward.

frac: numerators or denominators of sequence of rationals

Simple and useful.  I would like some way to formally tie the two 
sequences, so that
a program could generate the fractions.

full: the full sequence is given

Also straightforward.  A note that "fini" should also be specified 
would be useful.

hard: next term not known, may be hard to find. Would someone please 
extend this sequence?

Useful and sufficiently clear.  One might add a note that "more" need 
not also be specified.

more: more terms are needed! would someone please extend this sequence?

Useful and sufficiently clear.  There are a couple of problems.  One is 
that sometimes
someone will send in the additional terms, but the more keyword doesn't 
get removed.

I also think this keyword should be applied to any sequence with less 
than one full line of
values, unless there is a reason not to, because we want to have 
sufficient data to
disambiguate this from other sequences.  Reasons not to include: 
presence of "hard"
keyword; presence of "full" keyword; the next term is too big (in which 
there should be
a comment to that effect) - I believe there was once a "huge" keyword 
for this case,
but it is no longer used.

mult: Multiplicative: a(mn)=a(m)a(n) if g.c.d.(m,n)=1

Clear, and (now) pretty uniformly applied.

new: New (added within last two weeks, roughly)

Used by Neil.  Works.

nice: an exceptionally nice sequence

A judgement call.  Works better than "dumb", but it's very subjective.

nonn: a sequence of nonnegative numbers

Should specify that the terms present are nonnegative.

obsc: obscure, better description needed

Used by Neil.  Seems to perform its intended function.

sign: sequence contains negative numbers The %V, %W, %X lines (matching 
the %S,T,U lines) give the signed sequence

Same as for nonn.

tabf: An irregular (or funny-shaped) array of numbers made into a 
sequence by reading it row by row

This works.  I would like to have a formal way to specify the shape, so 
that these can
be automatically formatted.

tabl: typically a triangle of numbers, such as Pascal's triangle, made 
into a sequence by reading it row by row.

Generally works.

uned: Not edited. I normally edit all incoming sequences ...

Serves a useful function.

unkn: little is known; an unsolved problem; anyone who can find a 
formula or recurrence is urged to let me know.

Seems to be useful.

walk: counts walks (or self-avoiding paths)

This seems clear.  I think there are a fair number of sequences it 
applies to that don't
have it.

word: depends on words for the sequence in some language

Clear and useful, if only negatively.

Franklin T. Adams-Watters


-----Original Message-----
From: Antti Karttunen <antti.karttunen at gmail.com>

... 
Moreover, I think that on the whole the keyword-system of OEIS should 
be made 
more exact and extensive. E.g. any "hardcore combinatorial" sequence 
that is known to count some structures should contain keyword like 
"enum" 
or "count", permutations of integers keyword "perm", and so on. 
That way the serious users of the database wouldn't need to bother 
so much with all the "base" and "avant-garde" experimentations 
of us the tinkerers.