Based on the fact that $$\frac{1}{2}\ln\frac{1+x}{1-x}=\sum_{k=0}^\infty\frac{x^{2k+1}}{2k+1}$$ and evaluating at $x=-1/3$, we can conclude $$\ln 2=2\sum_{k=0}^\infty \frac{1}{(2k+1)3^{2k+1}}\,.$$

However, I'm interested in the speed of this convergence. Is it true that this series converges to the limit linearly in the sense that $$\limsup_{n\rightarrow\infty}\frac{|\ln 2-S_{n+1}|}{|\ln 2-S_{n}|}<1$$ where $\{S_n\}$ are the partial sums?

It's pretty straight-forward to calculate $$|\ln 2-S_n|=\sum_{k=n+1}^\infty\frac{1}{(2k+1)3^{2k+1}}<\frac{1}{2n+3}\sum_{k=n+1}^\infty\frac{1}{3^{2k+1}}=\frac{1}{8(6n+9)9^n}\,.$$ However, an effective lower bound on this tail is a little harder to find that'll make the ratio work out.


We can use integral remainders:

\begin{align}\sum_{k=n}^\infty\frac1{(2k+1)3^{2k+1}}&=\sum_{k=n}^\infty\int_0^{1/3}x^{2k}~\mathrm dx\\&\stackrel?=\int_0^{1/3}\sum_{k=n}^\infty x^{2k}~\mathrm dx\\&=\int_0^{1/3}\frac{x^{2n+1}}{1-x^2}~\mathrm dx\\&>\int_0^{1/3}\frac{x^{2n+1}}{1-(1/3)^2}~\mathrm dx\\&=\frac1{16(n+1)9^n}\end{align}

which should be sufficient. Note this also proves the slightly tighter upper bound:

$$\sum_{k=n}^\infty\frac1{(2k+1)3^{2k+1}}<\int_0^{1/3}\frac{x^{2n+1}}{1-0^2}~\mathrm dx=\frac1{18(n+1)9^n}$$

and hence you can see that


  • $\begingroup$ That makes me wonder for what $\alpha$ is the tail sequence equivalent to $\alpha/(n9^n)$. Different question though, and something else for me try. Thanks! $\endgroup$ – Robert Wolfe Jan 12 at 1:19
  • 1
    $\begingroup$ @RobertWolfe Hint: integrate my last non-bound integral, by parts. $\endgroup$ – Simply Beautiful Art Jan 12 at 1:20

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.