That's interesting I suppose...

Calculus Level 5

For recursive relation x n + 1 = x n 2 + 1 2 x n 1 x_{n+1} = \dfrac {x_n^2+1}{2x_n-1} , where n n is a positive integer,

lim n x n = { α if x 1 > 1 2 β if x 1 < 1 2 \lim_{n \to \infty} x_n = \begin{cases} \alpha & \text{if }x_1 > \frac 12 \\ \beta & \text{if }x_1 < \frac 12 \end{cases}

Find k = 0 ( α k β k ) 2 x k \displaystyle \sum_{k=0}^\infty \left(\alpha^k - \beta^k \right)^2 x^k for x = 0.1 x=0.1 .


The answer is 0.576.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

1 solution

Chew-Seong Cheong
Dec 11, 2019

Let's check the extrema of x n + 1 = x n 2 + 1 2 x n 1 x_{n+1} = \dfrac {x_n^2+1}{2x_n-1} .

d x n + 1 d x n = 2 x n ( 2 x n 1 ) 2 ( x n 2 + 1 ) ( 2 x n 1 ) 2 = 2 ( x n 2 x n 1 ) ( 2 x n 1 ) 2 \begin{aligned} \frac {d x_{n+1}}{dx_n} & = \frac {2x_n(2x_n-1)-2(x_n^2+1)}{(2x_n-1)^2} = \frac {2(x_n^2-x_n-1)}{(2x_n-1)^2} \end{aligned}

Implying that d x n + 1 d x n = 0 \dfrac {d x_{n+1}}{dx_n} = 0 , when x n = 1 ± 5 2 = { φ 1 φ x_n = \dfrac {1\pm \sqrt 5}2 = \begin{cases} \varphi \\ 1-\varphi \end{cases} , where φ \varphi is the gold ratio .

Since { x n > 0 for x 1 > 1 2 x n < 0 for x 1 < 1 2 { x n [ φ , ) for x 1 > 1 2 x n ( , 1 φ ] for x 1 < 1 2 \begin{cases} x_n > 0 & \text{for }x_1 > \frac 12 \\ x_n < 0 & \text{for }x_1 < \frac 12 \end{cases} \implies \begin{cases} x_n \in [\varphi, \infty) & \text{for }x_1 > \frac 12 \\ x_n \in (-\infty, 1-\varphi] & \text{for }x_1 < \frac 12 \end{cases} .

Since x n x_n is bounded, let's assume lim n x n \displaystyle \lim_{n \to \infty} x_n converges to a limit \ell , then

lim n x n = lim n x n + 1 = lim n x n 2 + 1 2 x n 1 Since lim n x n = = 2 + 1 2 1 2 1 = 0 = { α = φ if x 1 > 1 2 β = 1 φ if x 1 < 1 2 \begin{aligned} \lim_{n \to \infty} x_n & = \lim_{n \to \infty} x_{n+1} \\ & = \lim_{n \to \infty} \frac {x_n^2+1}{2x_n-1} & \small \blue{\text{Since }\lim_{n \to \infty} x_n = \ell} \\ \implies \ell & = \frac {\ell^2+1}{2\ell-1} \\ \ell^2 - \ell - 1 & = 0 \\ \ell & = \begin{cases} \alpha = \varphi & \text{if }x_1 > \frac 12 \\ \beta = 1 - \varphi & \text{if }x_1 < \frac 12 \end{cases} \end{aligned}

We get the same results by finding the extrema of x n + 1 x_{n+1} with x n = φ , 1 φ x_n = \varphi, 1-\varphi

{ For x n = φ , x n + 1 = φ 2 + 1 2 φ 1 = φ 2 + φ 2 φ 2 φ 1 = φ = x n For x n = 1 φ , x n + 1 = ( 1 φ ) 2 + 1 2 ( 1 φ ) 1 = 2 φ 1 φ 2 1 2 φ 1 = 1 φ = x n \begin{cases} \text{For }x_n = \varphi, & x_{n+1} = \dfrac {\varphi^2+\blue 1}{2\varphi -1} = \dfrac {\varphi^2+\blue{ \varphi^2-\varphi}}{2\varphi -1} = \varphi & = x_n \\ \text{For }x_n = 1-\varphi, & x_{n+1} = \dfrac {(1-\varphi)^2+1}{2(1-\varphi) -1} = \dfrac {2\varphi - 1 - \varphi^2-1}{2\varphi -1} = 1-\varphi & = x_n \end{cases}

This means that x n + 1 = x n x_{n+1} = x_n at the extrema of α = φ \alpha = \varphi and β = 1 φ \beta = 1-\varphi . That is x n x_n converges to α \alpha and β \beta .

Now consider k = 0 ( α k β k ) 2 x k \displaystyle \sum_{k=0}^\infty \left(\alpha^k - \beta^k\right)^2 x^k for x = 0.1 x=0.1 or

k = 0 ( φ k ( 1 φ ) k ) 2 1 0 k = k = 1 5 F k 2 1 0 k 0.567 where F n is the n th Fibonacci number. \begin{aligned} \sum_{k=0}^\infty \frac {\left(\varphi^k - (1-\varphi)^k\right)^2}{10^k} & = \sum_{k=1}^\infty \frac {5F_k^2}{10^k} \approx \boxed{0.567} & \small \blue{\text{where }F_n \text{ is the }n\text{th Fibonacci number.}} \end{aligned}


Reference: Fibonacci number

For those of you out there that may be wondering, the closed form for the generating function k = 0 F k 2 x k \sum_{k=0}^{∞} F_{k}^{2}x^{k} is x ( 1 x ) ( x + 1 ) ( 1 3 x + x 2 ) \frac{x(1-x)}{(x+1)(1-3x+x^{2})} :)

Zhang Xiaokang - 1 year, 6 months ago

1 pending report

Vote up reports you agree with

×

Problem Loading...

Note Loading...

Set Loading...