[seqfan] A071390

[israel at math.ubc.ca]

A071390 is "Least number m such that sigma(m) - phi(m) = n, or 0 if no such 
m exists."

The Mathematica code computes sigma(m) - phi(m) for m up to 1000,
and takes a(n)=0 if none of these is m.  I have no doubt that this
gives the correct values for the Data (up to n= 78), but it would
be nice to have a theoretical result for how far we need to search
in general.  For composite n it seems sigma(n)-phi(n) -> infinity
as n -> infinity (of course for primes the difference is 2).  Is 
there a convenient bound?

Cheers,
Robert