Subscripting (was: Base change notation)
Marc LeBrun
mlb at well.com
Thu Aug 17 01:00:57 CEST 2006
>=franktaw at netscape.net
> the usual convention with the underscore is that only the minimum
is considered to be subscripted
Seems reasonable. So for example 11_2 = 3 doesn't need any
subscripts. I'd supposed neither would 11_2->4, because the parsing
(11_2)->4 would be meaningless, but in more complex cases it might be
ambiguous, so we'll scope underscore so it "binds tightly".
> As for parens vs. braces, I think you can use either...Since we're
just presenting formulas in the OEIS...
But the OEIS is *not* just for presentation to humans, it's exposed
to programs. Having alternative synonyms tends to interfere with
that. Therefore the best practice to hew to some canonical pattern
whenever possible.
> (This applies to superscripts with ^, too.)
Ditto. Hm, so by analogy we can write 11_-1 = 1?
Mumble...now I'm remembering a reason why I avoided subscripts for
rebasing and instead again found "indexing" attractive: In the usual
usage of subscripts for bases the "main" term is treated as a
*string*, while the subscript is parsed in *decimal*.
I think it's very important, so as not to reintroduce all that
confusion, to find a way to get strings out of the picture
completely, so all arguments can be interpreted uniformly as ordinary
numbers and expressions.
For example, what's e_pi, if e_16 is fourteen?
And if you try to compose "subscripting" (besides getting squintier,
as with exponents), it gets considerably more confusing. It's not
clear what's a string, where the implied _10s go, etc.
So maybe bringing subscripting ideas into this isn't such a good idea
after all?
(I'll note that subscripting versus indexing is orthogonal to the
question of argument order--connecting the "from" and "to" bases by
an arrow certainly has its charms).
On a different note: no matter which of these notational proposals we
adopt, the OEIS convention of ignoring punctuation makes it hard to
find, much less change/maintain entries involving purely symbolic
notations. Maybe it would be better to introduce an operator with an
alphanumeric name, such as "rebase"?
More information about the SeqFan
mailing list