Mobius terminology
N. J. A. Sloane
njas at research.att.com
Sun Mar 20 18:02:38 CET 2005
On Mar 10 (while I was away) Emeric said:
> It seems that there is some confusion at some OEIS sequences
> regarding the Mobius transform.
> Given a function g(n), defined on the positive integers, the
> Mobius transform of g is
> f(n)=sum_{d|n} g(d)
> and then, of course, g is the inverse Mobius transform of f.
> Not the other way around.
> The number theory books I have do mention Mobius pairs but do
> not define the "Mobius transform".
Me: I respectfully disagree! In my "transforms" page, I give
this definition:
MOBIUS:=proc(a) local b,i,d:
if whattype(a) <> list then RETURN([]); fi:
b:=[]:
for i to nops(a) do b:=[op(b), add( mob(i,d)*a[d], d=1..i)]: od:
RETURN(b);
end:
which is equivalent to
f(n) = sum_{d|n} mobius(n/d) g(d)
This is consistent with the standard usage that the
K transform (for some kernel K) of a sequence {g} is the sequence {f}
given by an expression of the form
f(n) = Sum_{i} K(n,i) g(i)
The binomial transform, for example, is
f(n) = Sum_{i=0..n} C(n,i) g(i)
NJAS
More information about the SeqFan
mailing list