0

According to WolframAlpha partial sums for

http://www.wolframalpha.com/input/?i=sum%28prime%28n%29%2Fprime%28n%2B1%29%2Fn%5E2%2Cn%3D1...infinity%29&h=1

(I actually used the Maple notation for the input)

sum(prime(n)/prime(n+1)/n^2,n=1...infinity)

seems to converge value-wise (looking at partial sums) in the vicinity of 1.12984...

Are there any references to such calculations using actual (known by now) prime numbers?

Alex
  • 56

1 Answers1

1

Of course we have the trivial upper bound
$$ \sum_{n=1}^{\infty}\frac{p_n}{p_{n+1}n^2}<\sum_{n=1}^{\infty}\frac{1}{n^2}=\frac{\pi^2}{6}. $$ Hence the sum converges. For better estimates for the sum we could use $$ \log n+\log \log n-1<{\frac {p_{n}}{n}}<\log n+\log \log n\quad {\text{for }}n\geq 6. $$ or even better estimates, see here.

Dietrich Burde
  • 130,978