base keyword usage
Marc LeBrun
mlb at fxpt.com
Thu Jun 16 01:51:04 CEST 2005
>=Franklin T. Adams-Watters
> [the base keyword] should, in my opinion, mean "this is a sequence about
how numbers are written, not about mathematics".
I'm afraid in practice the meaning of "base" has become hopelessly muddled
and hard to change; this discussion is probably entirely moot.
So of course I'll comment (with a nod to Saint Jude)...
The keyword I'd *really* like to have is one that indicated "this sequence
fundamentally depends on that silly decimal representation" (or "decimal
encryption" as Bill Gosper says). Then I could filter out most of what I
consider irrelevant.
But that's just my taste, and tastes vary. For example NJAS has included
sequences like the "decimal expansion of pi", A00796.
Is that sequence "about how numbers are written", or is it "about mathematics"?
Of course base isn't just decimal, but any radix, thereby muddling things
further.
Thus I feel obligated to check the base box whenever I submit a sequence
that in any way involves binary expansions, though I may consider it
perfectly serious (as opposed to, say, my "dismal arithmetic"
contributions, although there's a serious general base-parameterized
concept behind even those).
It might help if instead of just a keyword there was a whole field for
specifying the base, which would be 10 for decimal, 2 for binary and so on.
But then there's "4[n]2" A065362, the "Zeckendorf expansion" A035514, the
"Fibbinary numbers" A003714, et al where more than one base is involved, or
exotic, or both.
Sometimes base tries to indicate that the sequence values are "in" some
base b, but when that's the case they can always equivalently be regarded
as ordinary values that could be expressed in rebase notation as b[a(n)]10.
Should base just indicate that some basic-expansion has something to do
with the sequence? If so then should it be checked for the
Moser-de Bruijn sequence A000695, because that sequence can be defined in a
basey way--as 2[n]4--or should it not be checked since there's a non-basey
alternative--"sums of distinct powers of 4"?
Finally, should we abuse base by further overloading or should there be
another keyword to signal "stringy" sequences such as the cellular
automaton A006977 or the "Look and Say" sequence A005150?
Nowadays on submission I just check it out of guilt whenever I can imagine
someone yelling at me later for not doing so. However I don't expect it to
convey anything precise enough to be useful on lookups.
More information about the SeqFan
mailing list