Golden Iterations II (updated)

Calculus Level 5

See Original Golden Iterations I

Note: This problem has been updated with notes from Chris and Mark. Thank you!

You can find the value of ϕ = 1 + 5 2 \phi=\dfrac{1+\sqrt{5}}{2} by iteration. Start with x 1 = 1 x_1=1 and x 2 = 1 + x 1 = 1.4142... x_2 = \sqrt{1+x_1}=1.4142... and for the ( n + 1 ) (n+1) th iteration:

x n + 1 = 1 + x n \large x_{n+1}=\sqrt{1+x_n}

For large n n , we can express the error term as e n = x n ϕ |e_n|=|x_n-\phi| . It can be shown that lim n e n + 1 e n d = b a c \displaystyle \lim_{n\rightarrow\infty}\frac{|e_{n+1}|}{|e_n|^d}=\frac{\sqrt{b}-a}{c} , where a , b , c a, b, c are integers that do not have common divisors. Find a + b + c + d a+b+c+d .

If you like this problem: Try out Golden Iterations III


The answer is 11.

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.

4 solutions

Chris Lewis
Jun 6, 2019

Write x n = φ + ε n x_n=\varphi+\varepsilon_n . (The " ε \varepsilon "s and " e e "s from the question are the same in absolute value, but it's useful to work with the error term before taking the absolute value).

We have x n + 1 = 1 + x n = 1 + φ + ε n x_{n+1}=\sqrt{1+x_n}=\sqrt{1+\varphi + \varepsilon_n}

Since φ 2 = 1 + φ \varphi^2=1+\varphi , this can be rewritten as x n + 1 = φ 1 + ε n 1 + φ x_{n+1} = \varphi\sqrt{1 + \frac{\varepsilon_n}{1+\varphi}}

Since for large n n , ε n 1 |\varepsilon_n| \ll 1 , we can apply the binomial theorem to get

x n + 1 = φ ( 1 + ε n 2 ( 1 + φ ) ) + O ( ε n 2 ) x_{n+1} = \varphi \left(1 + \frac{\varepsilon_n}{2(1+\varphi)} \right)+O(\varepsilon_n^2)

Subtracting φ \varphi from both sides we get

ε n + 1 = φ ε n 2 ( 1 + φ ) + O ( ε n 2 ) = ε n 2 φ + O ( ε n 2 ) \varepsilon_{n+1} = \frac{\varphi \varepsilon_n}{2(1+\varphi)} + O(\varepsilon_n^2)\\ =\frac{\varepsilon_n}{2\varphi} + O(\varepsilon_n^2)

So in the limit, we have

ε n + 1 ε n = 1 2 φ = φ 1 2 = 5 1 4 \frac{\varepsilon_{n+1}}{\varepsilon_n}=\frac{1}{2\varphi}=\frac{\varphi-1}{2}=\frac{\sqrt{5}-1}{4}

and finally ( a , b , c , d ) = ( 5 , 1 , 4 , 1 ) (a,b,c,d)=(5,1,4,1) giving the answer a + b + c + d = 11 a+b+c+d=11 .

Aareyan Manzoor
Jun 5, 2019

notice that lim n x n = ϕ \lim_{n\to \infty} x_n = \phi . let f ( x ) = 1 + x f(x)= \sqrt{1+x} , then the limit in the problem is the absolute value of lim n f ( x n ) ϕ ( x n ϕ ) d = lim x n ϕ f ( x n ) ϕ ( x n ϕ ) d \lim_{n \to \infty} \dfrac{f(x_n) - \phi}{(x_n-\phi)^d}= \lim_{x_n \to \phi} \dfrac{f(x_n) - \phi}{(x_n-\phi)^d} applying l'hopitals lim x n ϕ f ( x n ) ϕ ( x n ϕ ) d = lim x n ϕ f ( x n ) d ( x n ϕ ) d 1 \lim_{x_n \to \phi} \dfrac{f(x_n) - \phi}{(x_n-\phi)^d} = \lim_{x_n \to \phi} \dfrac{f'(x_n)}{d(x_n-\phi)^{d-1}} notice for d > 1 d>1 this diverges and for d < 1 d<1 it is 0 0 . hence d = 1 d=1 must be the solution, giving lim x n ϕ f ( x n ) 1 ( x n ϕ ) 1 1 = f ( ϕ ) = 1 2 1 + ϕ = 1 2 ϕ = 5 1 4 \lim_{x_n \to \phi} \dfrac{f'(x_n)}{1(x_n-\phi)^{1-1}} = f'(\phi) =- \dfrac{1}{2\sqrt{1+\phi}} = -\dfrac{1}{2\phi}\\ = - \boxed{\dfrac{\sqrt{5}-1}{4}}

K T
Jun 5, 2019

Observe that e n + 1 = 1 + x n φ = 1 + φ + e n φ e_{n+1}=\sqrt{1+x_n}-φ=\sqrt{1+φ+e_n}-φ . So we can express the limit in terms of e n e_n approaching 0.

lim e 0 1 + φ + e φ e d \lim_{e\to0} \frac{\sqrt{1+φ+e}-φ}{e^d}

Both numerator and denominator evaluate to 0 for e = 0 e=0 , so we use l'Hôpital's rule.

lim e 0 d d e 1 + φ + e φ d d e e d \lim_{e\to0} \frac{\frac{d}{de}\sqrt{1+φ+e}-φ}{ \frac{d}{de}e^d} = lim e 0 1 2 1 + φ + e d e d 1 =\lim_{e\to0}\frac{\frac{1}{2\sqrt{1+φ+e}}}{de^{d-1}} = lim e 0 1 2 d e d 1 1 + φ + e =\lim_{e\to0}\frac{1}{2de^{d-1}\sqrt{1+φ+e}}

This value is independent of e when we choose d=1. Our limit then evaluates to

lim e 0 1 2 1 + φ + e = 1 2 φ = 5 1 4 \lim_{e\to0} \frac{1}{2\sqrt{1+φ+e}} =\frac{1}{2φ}=\frac{\sqrt{5}-1}{4} .

So we have a = 1 , b = 5 , c = 4 , d = 1 a=1, b=5, c=4, d=1 and a + b + c + d = 11 a+b+c+d=\boxed{11}

Email Juegos
Jun 9, 2019

It is clear that lim n x n = ϕ \lim_{n \to \infty} x_n = \phi , so e n + 1 / e n d |e_{n+1}|/|e_n|^d will diverge except if the numerator cancels the terms of the form of x n ϕ x_n - \phi in the denominator.

We have

e n + 1 e n d = x n + 1 ϕ x n ϕ d . \displaystyle{\frac{|e_{n+1}|}{|e_n|^d} = \frac{|x_{n+1} - \phi|}{|x_n- \phi|^d} }.

We can rewrite the recursion of x n x_n to find

x n + 1 2 1 = x n . \displaystyle{ x_{n+1}^2 - 1 = x_n }.

Inserting this in the previous equation

e n + 1 e n d = x n + 1 ϕ x n + 1 2 1 ϕ d . \displaystyle{\frac{|e_{n+1}|}{|e_n|^d} = \frac{|x_{n+1} - \phi|}{|x_{n+1}^2 - 1 - \phi|^d} }.

Using that ϕ 2 = ϕ + 1 \phi^2 = \phi +1 , we get

e n + 1 e n d = x n + 1 ϕ x n + 1 2 ϕ 2 d = x n + 1 ϕ ( x n + 1 ϕ ) ( x n + 1 + ϕ ) d . \displaystyle{\frac{|e_{n+1}|}{|e_n|^d} = \frac{|x_{n+1} - \phi|}{|x_{n+1}^2 - \phi^2|^d} =\frac{|x_{n+1} - \phi|}{|(x_{n+1} - \phi) (x_{n+1} + \phi) |^d} }.

Only if d = 1 d =1 will we prevent the divergence in the limit. The result will then be

lim n e n + 1 e n d = lim n 1 ( x n + 1 + ϕ ) = 1 2 ϕ = 5 1 4 . \displaystyle{\lim_{n\to \infty} \frac{|e_{n+1}|}{|e_n|^d} = \lim_{n\to \infty} \frac{1}{| (x_{n+1} + \phi) |} = \frac{1}{2\phi} = \frac{\sqrt{5}-1}{4} }.

And so the answer is 5 + 1 + 4 + 1 = 11 5+1+4+1 = \boxed{11} .

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...