Repeating Radicals

Algebra Level 4

1 + 1 1 1 1 + 1 1 \large \sqrt{1 + \dfrac{1}{\sqrt{1 - \dfrac{1}{\sqrt{1 + \dfrac{1}{\sqrt{1 - \cdots}}}}}}}

If the above expression equals x , x, then determine 1000 x . \lfloor 1000x \rfloor.


The answer is 1618.

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

Let y y be similar to the given expression x x but with the addition and subtraction signs switched. Then we have that

x = 1 + 1 y x = \sqrt{1 + \dfrac{1}{y}} and y = 1 1 x y = \sqrt{1 - \dfrac{1}{x}}

x 2 y 2 = 1 y + 1 x ( x y ) ( x + y ) = x + y x y x y = 1 x y , \Longrightarrow x^{2} - y^{2} = \dfrac{1}{y} + \dfrac{1}{x} \Longrightarrow (x - y)(x + y) = \dfrac{x + y}{xy} \Longrightarrow x - y = \dfrac{1}{xy},

since clearly x + y 0 , x + y \ne 0, (as both x , y > 0 x,y \gt 0 ). We then have that

( x y ) x y = 1 x 2 y x y 2 = 1 (x - y)xy = 1 \Longrightarrow x^{2}y - xy^{2} = 1

( 1 + 1 y ) y x ( 1 1 x ) = 1 y + 1 ( x 1 ) = 1 x y = 1 \Longrightarrow \left(1 + \dfrac{1}{y}\right)y - x\left(1 - \dfrac{1}{x}\right) = 1 \Longrightarrow y + 1 - (x - 1) = 1 \Longrightarrow x - y = 1

x y = 1 x ( x 1 ) = 1 x 2 x 1 = 0 \Longrightarrow xy = 1 \Longrightarrow x(x - 1) = 1 \Longrightarrow x^{2} - x - 1 = 0

x = 1 + 5 2 = 1.618033989..... , \Longrightarrow x = \dfrac{1 + \sqrt{5}}{2} = 1.618033989....., since x > 0. x \gt 0.

Thus 1000 x = 1618 . \lfloor 1000x \rfloor = \boxed{1618}.

Brilliant ! :)

Yuki Kuriyama - 5 years, 9 months ago

Are we not worried about convergence?

Otto Bretscher - 5 years, 9 months ago

Log in to reply

Yes, we're always worried about convergence, but I wasn't sure how to go about proving convergence in the case of "entwined" sequences. I did plug about 10 iterations of the radical into WolframAlpha and the point of attraction did seem to be ϕ , \phi, at which point my interest in a formal proof waned; that is, until you brought the matter up again. I'll give it another go after doing a bit of research first.

Brian Charlesworth - 5 years, 9 months ago

Log in to reply

The way to show convergence is conceptually pretty straightforward, but the details are a bit tedious in this case, as you say. Our iteration function is f ( x ) = 1 + 1 1 1 / x f(x)=\sqrt{1+\frac{1}{\sqrt{1-1/x}}} for x > 1 x>1 ; this function is decreasing and concave up, with f ( 1.4 ) > 0.5 f'(1.4)>-0.5 . If we let a 1 = 2 a_1=\sqrt{2} and a n + 1 = f ( a n ) a_{n+1}=f(a_n) , then a n + 1 ϕ < 1 2 a n ϕ |a_{n+1}-\phi|<\frac{1}{2}|a_n-\phi| , by the mean value theorem, so that a n a_n does indeed converge to ϕ \phi .

Otto Bretscher - 5 years, 9 months ago

Log in to reply

@Otto Bretscher Interesting continued fraction series for Phi, the golden mean !!. I had a feeling that since this involves an infinite series of nested radicals, the solution will tend to a value close to 1 (which is not too far off in this case) and hence we need not be bothered about convergence. Also, the series consists of alternating sign expressions which further reinforces the above

Sundar R - 5 years, 8 months ago

@Otto Bretscher Great! Thanks for outlining the proof. :)

Brian Charlesworth - 5 years, 9 months ago

@Otto Bretscher How did you get this: a n + 1 ϕ < 1 2 a n ϕ |a_{n+1}-\phi|<\frac{1}{2}|a_n-\phi| ?

Vighnesh Raut - 5 years, 8 months ago

Log in to reply

@Vighnesh Raut By the mean value theorem we have f ( a n ) f ( ϕ ) a n ϕ = a n + 1 ϕ a n ϕ = f ( c ) \frac{f(a_n)-f(\phi)}{a_n-\phi}=\frac{a_{n+1}-\phi}{a_n-\phi}=f'(c) for some c c in between; we know that f ( c ) < 0.5 |f'(c)|<0.5 in the relevant domain.

Otto Bretscher - 5 years, 8 months ago

Log in to reply

@Otto Bretscher Oh! Thanks. Got it.

Vighnesh Raut - 5 years, 8 months ago

I came across similar type of problems but every time used different substitutions at beginning......from next time, i will keep this solution in mind..

Vighnesh Raut - 5 years, 9 months ago

Excellent solution different than other similar looking problems.

Shyambhu Mukherjee - 5 years, 8 months ago

Thanks for the simple solution.

Niranjan Khanderia - 5 years, 3 months ago

same method

Figel Ilham - 5 years, 9 months ago

in last line you have declared, x>0 why, x>0 ? can -618 be a value ?

Sanjoy Roy - 5 years, 8 months ago

same method

Mas Mus - 5 years, 8 months ago

x^2 * y = y+1 and y^2 * x = x-1 jus add

Razine Mohammed - 5 years, 8 months ago

I think I could make this problem even more scary looking. What if, instead of just 1's, it is 1, 2, 3, 4, ... in the fractions and roots.

Sharky Kesa - 5 years, 7 months ago

Log in to reply

That would indeed be scary; just in time for Halloween. We might have to invoke the spirit of Ramanujan to solve that radical. :)

Brian Charlesworth - 5 years, 7 months ago

Log in to reply

Yes, but it would be quite dangerous. How would one even attempt to solve it?

Sharky Kesa - 5 years, 7 months ago

Log in to reply

@Sharky Kesa Sorcery!? You're right, it would be dangerous; one could get trapped in it for days. Probably best to just whistle as you walk by it, pretending it's not there. :) But seriously though, while fun to think about, I doubt that it has a closed-form solution. Assuming that we are replacing the 1 1 's on the "diagonal" with 1 , 2 , 3 , 4 , 5 , . . . . . 1,2,3,4,5,..... it looks to approach about 1.351 , 1.351, which doesn't look that special.

Brian Charlesworth - 5 years, 7 months ago

Log in to reply

@Brian Charlesworth No, I mean something even more horrific:

1 + 2 3 4 5 + 6 7 8 9 + 10 11 . . . . \large \sqrt{1 + \dfrac{2}{\sqrt{3 - \dfrac{4}{\sqrt{5 + \dfrac{6}{\sqrt{7 - \dfrac{8}{\sqrt{9 + \dfrac{10}{\sqrt{11 - ....}}}}}}}}}}}

Sharky Kesa - 5 years, 7 months ago

@Brian Charlesworth Also, could it be the plastic constant?

x 3 = x + 1 x^3=x+1

Sharky Kesa - 5 years, 7 months ago

Whoa!!!!!!!!!

Hari prasad Varadarajan - 4 years, 5 months ago

What about the other solution x=-1.324, I don't see any constraint for x>0

Harout G. Vartanian - 4 years, 4 months ago

Log in to reply

By protocol, square roots are always taken to be non-negative. Therefore, we cannot have a negative answer.

Sharky Kesa - 4 years, 4 months ago
汶良 林
Sep 8, 2015

How did you see that a factorization was possible at all!!

Puneet Pinku - 4 years, 9 months ago

I also did the same.

Akhil Bansal - 5 years, 9 months ago

Inserting x was twist !great!

Shyambhu Mukherjee - 5 years, 8 months ago

After step 4 I used Wolfram. Your factorization is very elegant!

Indranil Chakraborty - 5 years, 8 months ago

cant x be a negative value? if not, what is the reason for that ?

Sanjoy Roy - 5 years, 8 months ago

Log in to reply

汶良 林 - 5 years, 8 months ago

Log in to reply

oow, such a easy explanation. thanks mate :)

Sanjoy Roy - 5 years, 8 months ago

Did the same :D

A Former Brilliant Member - 4 years, 10 months ago

I do not see any constraint for x>1 in the problem statement. So what about the other solution x=-1.3247?

Harout G. Vartanian - 4 years, 4 months ago

Log in to reply

since x is equal to a square root, it has to be > 0.

David Ortiz - 3 years, 7 months ago

Very tidy. I missed this simpler factorisation and solved by inspection and guesswork - knowing the solution lay within a range [1600,1625] which suggested the familiar answer.

Malcolm Rich - 4 years, 3 months ago

You factorised the equation into ( x 2 x 1 ) ( x 3 x + 1 ) = 0 (x^2-x-1)(x^3-x+1)=0 but then solved for x 3 x 1 = 0 x^3-x-1=0 . Why did you do this? None of the solutions to x 3 x + 1 = 0 x^3-x+1=0 are real.

Nathaniel Bronson - 2 years, 5 months ago

Wow! Very nice!

Pi Han Goh - 5 years, 9 months ago

Here is a dirty little solution:

y = 1 + 1 1 1 y y=\sqrt{1+\frac1{\sqrt{{1-\frac1y}}}} Which when rearranging gets us y = ( y 1 ) 3 ( y + 1 ) 2 y=(y-1)^3(y+1)^2 Let u = y 1 u=y-1 then u + 1 = u 3 ( u + 2 ) 2 0 = u 5 + 4 u 4 + 4 u 3 u 1 u+1=u^3(u+2)^2\\ \implies 0=u^5 + 4u^4 + 4u^3-u-1 Now Setting up a Newton-Raphson approximation: x n + 1 = x n x n 5 + 4 x n 4 + 4 x n 3 x n 1 5 x n 4 + 16 x n 3 + 12 x n 2 1 x_{n+1}=x_n-\frac{x_n^5+4x_n^4+4x_n^3-x_n-1}{5x_n^4+16x_n^3+12x_n^2-1} Iterate until about n = 4 n=4 where it should stop changing digits after 3 d.p. Then add one and you should get an answer of 1618 \boxed{1618} .

This is under the assumption that a calculator is allowed, which I take as implied due to the nature of the answer.

what is Newton-Raphson approximation

Akash singh - 4 years, 5 months ago

I use for this equation WolframAlpha -;).

Yuriy Kazakov - 3 years, 9 months ago

Mathematica does it in one line:

Floor[1000 x] /. Solve[Sqrt[1+1/Sqrt[1-1/x]]==x,{x}]

-a fool with a tool...

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...