Mobius aMUsements
Marc LeBrun
mlb at well.com
Sat May 5 08:49:13 CEST 2001
Recently NJAS somehow inspired me to play at the ultra-simple minded game
of negating all the primes in various constructions.
For example if you negate the primes in the prime factorization of n you get
A61019: 1,-2,-3,4,-5,6,-7,-8,9,10,-11,-12,-13,14,15,16,-17,-18,-19,...
which turns out to be the inverse Mobius transform of
A1615: 1,3,4,6,6,12,8,12,12,18,12,24,14,24,24,24,18,36,20,36,32,36,...
This sequence is described in the EIS as "Sublattices of index n in generic
2-dimensional lattice; also index of GAMMA_0(n) in SL(2,z)." (Anyone care
to roughly decipher this definition?)
Similarly you can negate the primes in the classic sigma sum-of-divisors
function, and get this new sequence of signed sums
A61020: 1,1,2,3,4,2,6,5,7,4,10,6,12,6,8,11,16,7,18,12,12,10,22,10,...
and so forth.
This led to noticing that there is another sequence transform/inverse pair
much like the Mobius transform, whose weights are "factors" of Mobius mu:
Transform: Weight function:
Mobius -1 to sum of exponents in prime factorization, except 0 when n
is squarefull
MobiusInv always +1, no exceptions
Maybeus -1 to sum of exponents in prime factorization, no exceptions
MaybeusInv always +1, except 0 when n is squarefull
For example MaybeusInv of 1,2,3,... gives us A1615 again, and applied to
1,1,1,... seems to be
A3444: 1,4,12,43,143,504,1768,6310,22610,81752,297160,1086601...
inscrutably described just as "Dissections of a polygon." (How so?)
Questions:
Are these transforms morally independent of the regular Mobius pair, or are
they trivially related in some way? (NJAS called mu "canonical"; can we
really sing from a different hymnal?)
Generatingfunctionologically speaking, the Mobius transforms connect power
series with Lambert series. What sort of expansions are connected by the
new pair?
Do these factored-mu weight functions appear anywhere? If so, by what
names? (I've been calling them nu and psi).
If there's anything to this I'm thinking it might be fun to apply these
transforms to the whole EIS corpus and see what crosslinks. Some of the
"hits" above encourage this, but the multiple appearance of A1615 might be
a hint that maybe this isn't really worthwhile, because I'm overlooking
some essential triviality. So any additional insight you can provide would
be most helpful...
Thanks!
More information about the SeqFan
mailing list