As stated in the title I am trying to prove $g_{n}=\sum_{k=1}^{2^n}\frac{2^{n}}{k}\chi_{\left(\left(\frac{k-1}{2^{n}}\right)^{2},\left(\frac{k}{2^{n}}\right)^{2}\right]}$ converges pointwise to $\frac{1}{\sqrt{x}}$ on $(0,1]$. Obviously $\left\lbrace\left(\left(\frac{k-1}{2^{n}}\right)^{2},\left(\frac{k}{2^{n}}\right)^{2}\right]\right\rbrace_{k=1}^{2^n}$ is a partition of the interval $(0,1)$ so for some $k\in\left\lbrace 1,2,\dots,2^n\right\rbrace$, $g_n(x) =\frac{2^n}{k}$.

I was thinking about taking the largest $n$ such that for some $k\in\left\lbrace 1,2,\dots,2^n\right\rbrace$, $x\in \left(\left(\dfrac{k-1}{2^n} \right)^2,\left(\dfrac{k}{2^n}\right)^2 \right)$ and then breaking that up to find a $N$ such that $x=\left(\dfrac{k}{2^N}\right)^2$, but I am not sure such a $n$ exists. Any help is welcome. I tried squeeze theorem as well, but I failed to find one of the bounds.

  • $\begingroup$ - Your sentence "obviously .. so for some $k\in\dots, g_n(x) = \frac {2^n}k.$" doesn't make sense to me, I don't know what you're trying to say. Probably because $x$ appears out of nowhere. - is it $\sum_{k=1}^{2n}$ or $\sum_{k=1}^{2^n}$? - There is no largest $n$ such that $x$ belongs to that interval (I suppose you mean $(a,b]$ instead of $(a,b)$? This is because the endpoints $k/2^n$ cover the entire interval $[0,1]$, so for each $n$, there's always a $k$. $\endgroup$ – Calvin Khor May 8 at 6:07
  • $\begingroup$ $\sum_{k=1}^{2^n}$ $\endgroup$ – Michael Cook May 8 at 6:09

For any $n$ and $x$ there is a unique $k_n$ such that $(\frac {k_n-1} {2^{n}})^{2} <x \leq (\frac {k_n} {2^{n}})^{2}$. Note that $\frac {k_n-1} {2^{n}} <\sqrt x \leq \frac {k_n} {2^{n}}$. This implies that $\frac {k_n} {2^{n}} \to \sqrt x$. Hence $g_n(x)=\frac {2^{n}} {k_n} \to \frac 1 {\sqrt x}$.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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