31 Oct
2015
31 Oct
'15
6:29 p.m.
On 2015-10-31 17:23, Warren D Smith wrote:
Gosper 1974 found a nice sum which is equal to H(n) = SUM(k=1..n) 1/k but "converges faster" and also interpolates it for non-integer n.
Question: Is there an analogous nice way to handle the "odd harmonic sum" 1/1 + 1/3 + 1/5 + ... + 1/(2n-1) ? Sure, just start n at 1/2 instead of 1. --rwg