Summation of Odd Reciprocals in terms of Harmonic Numbers
Jump to navigation
Jump to search
Theorem
- $\ds \sum_{k \mathop = 1}^n \dfrac 1 {2 k - 1} = H_{2 n} - \dfrac {H_n} 2$
where $H_n$ denotes the $n$th harmonic number.
Proof
\(\ds \sum_{k \mathop = 1}^n \dfrac 1 {2 k - 1}\) | \(=\) | \(\ds \sum_{k \mathop = 1}^n \dfrac 1 {2 k - 1} + \sum_{k \mathop = 1}^n \dfrac 1 {2 k} - \sum_{k \mathop = 1}^n \dfrac 1 {2 k}\) | ||||||||||||
\(\ds \) | \(=\) | \(\ds \sum_{k \mathop = 1}^{2 n} \dfrac 1 k - \dfrac 1 2 \sum_{k \mathop = 1}^n \dfrac 1 k\) | ||||||||||||
\(\ds \) | \(=\) | \(\ds H_{2 n} - \dfrac {H_n} 2\) |
$\blacksquare$
Historical Note
Donald E. Knuth originally published the following result in The Art of Computer Programming: Volume 1: Fundamental Algorithms, 3rd ed. ($1997$) as a solution to the exercise to evaluate $\ds \sum_{k \mathop = 1}^n \dfrac 1 {2 k - 1}$:
- $\ds \sum_{k \mathop = 1}^n \dfrac 1 {2 k - 1} = H_{2 n - 1} - \dfrac {H_{n - 1} } 2$
The online errata sheet, found via https://www-cs-faculty.stanford.edu/~knuth/taocp.html, offers the simpler result as a replacement.
Sources
- 1997: Donald E. Knuth: The Art of Computer Programming: Volume 1: Fundamental Algorithms (3rd ed.) ... (previous) ... (next): $\S 1.2.7$: Harmonic Numbers: Exercise $16$