7

I thought up this question recently, and I think I've figured out the partial sum: $$ S_n := \left(n\prod_{k=2}^{n-1} k^{2^{n-k}}\right)^{2^{-k}}. $$ But I don't even quite know if I'm on the right track. If I am, how do I find the limit of the above equation, and if not, how can I find it another way? Thanks.

Hakim
  • 10,311
undo_all
  • 255
  • 8
    Isn't it $1^{1/2}2^{1/4}3^{1/8}...n^{1/2^{n-1}}$? We could take logarithms to get $\sum \frac{\ln(k)}{2^k}$, which Mathematica sums as $-\text{PolyLog}^{(1, 0)}(0, 1/2)$. So the product goes to $e^{-\text{PolyLog}^{(1, 0)}(0, 1/2)}$. –  Apr 08 '14 at 18:55
  • @Jacobo: Actually, I think the general term is $n^{\frac{1}{2^n}}.$ – Geoff Robinson Apr 08 '14 at 18:59
  • @GeoffRobinson Yes, that is what is in the sum. I was thinking in writing less terms when I wrote it. –  Apr 08 '14 at 19:01

3 Answers3

7

This is Somos' quadratic recurrence constant $($see also$)$, whose value is about $1.661688^{^-}$ and which is not yet known to possess a closed form. For a similar expression, see the nested radical constant. I think this should help out.

Lucian
  • 49,312
  • 1
    You beat me to it! I computed the value for 100 iterations and googled the result. I offer the companion link http://www.doiserbia.nb.rs/img/doi/1452-8630/2010%20OnLine-First/1452-86301000029N.pdf – MPW Apr 08 '14 at 19:25
6

We see that $$ S_n = 1^{\frac{1}{2}}2^{\frac{1}{4}}3^{\frac{1}{8}}\ldots = \prod_{k=1}^{n}k^{2^{-k}}. $$ Furthermore $$ \log S_n = \sum_{k=1}^{n}\log k^{2^{-k}} = \sum_{k=1}^{n}\left(\frac{1}{2}\right)^k\log k. $$ So we see that $$ \lim_{n\rightarrow\infty}\log S_n = \sum_{k=1}^{\infty}\left(\frac{1}{2}\right)^k\log k \le \sum_{k=1}^{\infty}k\left(\frac{1}{2}\right)^k = \\\frac{1}{2}\sum_{k=1}^{\infty}k\left(\frac{1}{2}\right)^{k-1} = \frac{1}{2}\frac{1}{(1-1/2)^2} = \frac{1}{8}<\infty. $$ So the product definitely converges. Unfortunately I don't know about the exact limit.

Marc
  • 6,856
0

An easy way of showing that the product is bounded, is to replace the sequence with $2^n$ instead of $n$.

The product is less that $ 2^{\frac{1}{2}} \times 2^{\frac{2}{4} } \times 2^{ \frac{3}{8} } \times 2^{ \frac{ 4} { 16} } \ldots = 2^2 = 4$.

Calvin Lin
  • 77,541