18

Context

I found an extra-hard (but fake) example of a paper which adheres to the syllabus for the hardest mathematics high school course available in the state of NSW in Australia; Mathematics Extension 2. Question 16 (a) has caught my eye and I'm having difficulty solving it, so am looking for tips. I will reproduce the question entirely below and add explain where I am up to.

Question Statement

$16 (a)$ $[\mathbf7$ $\mathbf{marks}]$ A sequence $x_1, x_2, x_3, \dots$ of positive real numbers is given, recursively defined by $x_1=1$ and $x_{n+1} = \frac{1}{x_1^2 + x_2^2 + \dots + x_{n}^2}$ for each $n > 1$.

$(i)$ $[\mathbf2]$ Show that $x_n \leq \frac{1}{\sqrt[3]{n-1}}$ for each $n\geq2$.

$(ii)$ $[\mathbf1]$ Prove that every integer $n\geq2$ satisfies $\frac{1}{2^{2/3}} + \frac{1}{3^{2/3}} + \dots + \frac{1}{n^{2/3}} \leq \int_1^nx^{-2/3}dx$, and hence show that $x_{n+1} > \frac{1}{3\sqrt[3]{n-1}}$ for each integer $n\geq2$.

$(iii)$ $[\mathbf1]$ Prove that every $n\geq2$ also satisfies $\frac{1}{\sqrt[3]{1}} + \frac{1}{\sqrt[3]{2}} + \dots + \frac{1}{\sqrt[3]{n}} \geq \int_1^{n+1}\frac{1}{\sqrt[3]{x}}dx$, and hence deduce that $x_1 + x_2 + \dots + x_n > \frac{(n-1)^{2/3}}{2}$ for each $n\geq2$.

$(iv)$ $[\mathbf1]$ By considering a similar integral in $(iii)$, show that for all sufficiently large positive integers $n$ the inequality $\frac{1 - \frac{1}{2019}}{2} < \frac{x_1 + x_2 + \dots + x_n}{n^{2/3}} < \frac{3 + \frac{1}{2019}}{2}$ holds.

$(v)$ $[\mathbf2]$ It is given ($\mathbf{Do}$ $\mathbf{NOT}$ $\mathbf{prove}$) that the limit $\lim\limits_{n \to \infty} \frac{x_1 + x_2 + \dots + x_n}{n^{2/3}}$ exists, and is equal to a positive real number $C$.

Show that $C = \frac{\sqrt[3]{9}}{2}$.

Attempt so far

I was able to solve parts (i), (ii) and (iii) and am able to solve the lower bound in (iv) but cannot find a good integral for the upper bound. (My problem is trying to make n+1 appear in the integrand, while also only using n Riemann sums for the bound). I also see issues with applying squeeze law in (v).

Request For Hint

What is a good hint for finding the correct integral in (iv), and how do I extend this to part (v) where I'm hoping squeeze law applies but can't see a trivial path forward if the two integral bounds have multipliers of $\frac{1}{2}$ and $\frac{3}{2}$ respectively (thereby squeezing to a range rather than a point).

Working for part (iii)

Page 1 of working

And an asymptotic hack to get out part (iv)

Page 2 of working

Gary
  • 36,640
  • 1
    Can you sho your work for $(iii)$? For $(iv)$, you do the same thing, but use slightly different rectangles. In $(i), (ii)$, you showed that $ 1/(3\sqrt[3]{n-2} ) < x_n < 1/\sqrt[3]{n-1}$, which helps explain where the factor of 3 comes from. – Calvin Lin Feb 03 '24 at 07:30
  • on that note, how to do Q11a. – sun_Jiaoliao Feb 03 '24 at 08:22
  • @CalvinLin I've added some working for part (iii) and shown a disgusting asymptotic hack to get out part (iv) (upper bound only, but lower bound is clear from iii), but I think the method I've used is too complicated for what high schoolers should use. – Juan Reight Deag Feb 03 '24 at 08:41
  • As a follow up, I found a solution to part (v) using Stolz–Cesàro theorem. I'd never heard of this theorem before, but you can do something nice with this problem setup (and the knowledge that the limit exists a priori) to get it out.

    I'm wondering if there's a more elementary method though, as parts (i)-(iv) were not needed to use the theorem; and perhaps there's a nicer thing there.

    – Juan Reight Deag Feb 03 '24 at 10:39
  • 1
    @JuanReightDeag The paper is called a "troll" paper, and my suspicion from looking at the other questions is that the last part is independent / requires much more machinery than the previous leaders. Stolz-Cesaro was also what I thought of, though isn't really appropriate for the age group here. – Calvin Lin Feb 03 '24 at 17:30
  • 1
    @JuanReightDeag FYI Your diagram for (iii) is wrong. You are using left-hand rectangles and overestimating. $\quad$ In fact, you drew the diagram for (iv), where we want to use right-hand rectangles and underestimate. – Calvin Lin Feb 03 '24 at 17:33
  • @CalvinLin sorry about that. I should have made it more clear, those are the triangles for question $(iv)$. I apologise about my poorly delimited work. Left of $Also$ should be $(iv)$, it's the start of work for (iv).

    Were there better Riemann sums (or otherwise) for (iv). I had to go back to asymptotic analysis to demonstrate that the extra constant bit was sufficient (which is a little too advanced). For the lower bound you can just directly solve for n using $\frac{(n-1)^{2/3}}{2n^{2/3}} = \frac{1 - \frac{1}{2019}}{2}$, which is a high school technique.

    Maybe that's the (troll) point!

    – Juan Reight Deag Feb 03 '24 at 21:48
  • Please do not use pictures for critical portions of your post. Pictures may not be legible, cannot be searched and are not view-able to some, such as those who use screen readers. – Martin R Feb 05 '24 at 14:50

1 Answers1

4

This is my first time answering question on this website.If my editing form is wrong,please forgive me.Thanks. $$ x_{n+1}=\frac{1}{{x_1}^2+{x_2}^2+...+{x_{n-1}}^2+{x_n}^2} \\ so\,\,x_n=\frac{1}{{x_1}^2+{x_2}^2+...+{x_{n-1}}^2}\,\,and\,\,\frac{1}{x_n}={x_1}^2+{x_2}^2+...+{x_{n-1}}^2 \\ so\,\,x_{n+1}=\frac{1}{\frac{1}{x_n}+{x_n}^2}=\frac{x_n}{1+{x_n}^3}\,\,We\,\,got\,\,the\,\,recurrence\,\,relation. \\ \,\,x_1=1 ,according\,\,to\,\,recurrence\,\,relation, we\,\,know\,\,x_{n+1}-x_n<0, x_n\rightarrow 0\left( n\rightarrow \infty \right) \\ \frac{1}{{x_{n+1}}^3}=\frac{1+3{x_n}^3+3{x_n}^6+{x_n}^9}{{x_n}^3}=\frac{1}{{x_n}^3}+3+O\left( {x_n}^3 \right) \\ so\,\, \frac{1}{{x_{n+1}}^3}-\frac{1}{{x_n}^3}=3+O\left( {x_n}^3 \right) \\ so\,\,\frac{1}{{x_{n+1}}^3}-\frac{1}{{x_1}^3}=3n+\sum_{n=1}^n{O\left( {x_n}^3 \right)} \\ \frac{1}{{x_{n+1}}^3}=3n+\sum_{n=1}^n{O\left( {x_n}^3 \right)}+\frac{1}{{x_1}^3} \\ x_{n+1}\sim \left( \frac{1}{3n} \right) ^{\frac{1}{3}} \\ \lim_{n\rightarrow \infty} \frac{x_1+x_2+...x_{n+1}}{n^{\frac{2}{3}}}\overset{OStolz\,\,therom}{=}\lim_{n\rightarrow \infty} \frac{x_{n+1}}{\left( n+1 \right) ^{\frac{2}{3}}-n^{\frac{2}{3}}}=\lim_{n\rightarrow \infty} \frac{\left( \frac{1}{3n} \right) ^{\frac{1}{3}}}{n^{\frac{2}{3}}\left( \left( 1+\frac{1}{n} \right) ^{\frac{2}{3}}-1 \right)}=\lim_{n\rightarrow \infty} \frac{\left( \frac{1}{3n} \right) ^{\frac{1}{3}}}{n^{\frac{2}{3}}\frac{2}{3}\frac{1}{n}}=\frac{\sqrt[3]{9}}{2} $$

megumin
  • 41
  • As it’s currently written, your answer is unclear. Please [edit] to add additional details that will help others understand how this addresses the question asked. You can find more information on how to write good answers in the help center. – Community Feb 07 '24 at 18:35