Variation On Harmonic Series Approximates Log

[Edwin Clark]

On Sun, 13 Jul 2008, Robert Israel wrote:

> I get
> 0, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
> 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1
> ...
>
> where a[3], a[12] and a[164] are 2, and all the other a[n] for 2 <= n <= 4000 
> are 1.


I wrote a Maple program that computes the sequence to n = 132,000. My 
results agree with Robert's to n = 4000. Further I get a[49733] = 2 and
the rest 1's when n > 1. Maybe someone could check the value of 
a[49733]?  I set the value of Digits to 500, but I am a little leery when 
summing such a series.

Unfortunately the sequence 3, 12, 164, 49733 doesn't appear in the OEIS.

Edwin