Inspired by John Muradeli

Calculus Level 3

lim n cos cos cos cos n x \lim_{n \rightarrow \infty} \underbrace{\cos \cos \cos \ldots \cos}_{n} \ x

Evaluate the limit above. Give your answer rounded to three decimal places.

If you think the limit does not exist, submit your answer as 123.


The answer is 0.739.

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.

3 solutions

Arturo Presa
Oct 10, 2015

Let x x be a an arbitrary real number and ( x n ) n (x_{n})_n be a sequence defined by the recursion x n + 1 = cos x n , x_{n+1}=\cos x_{n}, where x 1 = cos x x_{1}=\cos x . Obviously, we have that x n 1 |x_{n}| \leq 1 for any natural number n n . On the other hand, using the mean value theorem , we have that cos a cos b = sin ( c ) ( a b ) \cos a- \cos b = \sin (c) (a-b) for a certain number c c in between a a and b b . Therefore if a a and b b are in the interval [ 1 , 1 ] [-1, 1] the following inequality holds cos a cos b r a b ( ) |\cos a -\cos b|\leq r|a-b|\:\:\:\:\:\:(*) where r = sin ( 1 ) r= \sin(1) and it is obvious that 0 < r < 1 0<r<1 . Now, using mathematical induction we can prove that x n + 1 x n r n 1 x 2 x 1 . |x_{n+1}-x_{n}|\leq r^{n-1}| x_{2}-x_{1}|. Indeed, this is true for n = 1 n=1 and now if we assume the inequality is true when n = k , n=k, then using ( ) (*) we obtain that x k + 2 x k + 1 = cos x k + 1 cos x k r x k + 1 x k r k x 2 x 1 . |x_{k+2}-x_{k+1}|=|\cos x_{k+1}-\cos x_{k}|\leq r |x_{k+1}-x_{k}|\leq r^{k}|x_{2}-x_{1}|.

Now, we can also see the following: x n + m x n x n + m x n + m 1 + . . . + x n + 2 x n + 1 + x n + 1 x n ( r n + m 2 + . . . + r n 1 ) x 2 x 1 |x_{n+m}-x_{n}|\leq |x_{n+m}-x_{n+m-1}|+...+|x_{n+2}-x_{n+1}|+|x_{n+1}-x_{n}|\leq (r^{n+m-2}+...+r^{n-1})|x_{2}-x_{1}| r n 1 ( r m 1 + r m 2 + . . . + 1 ) x 2 x 1 < r n 1 1 r x 2 x 1 . \leq r^{n-1}(r^{m-1}+r^{m-2}+...+1)|x_{2}-x_{1}|<\frac{r^{n-1}}{1-r}|x_{2}-x_{1}|. The latter inequality implies that x n x_{n} is a Cauchy sequence and therefore is convergent. So lim n x n \lim_{n \to \infty}{x_{n}} exists and is a finite number that we can denote by l l . Now taking limits in both sides of the equations x n + 1 = cos x n , x_{n+1}=\cos x_{n}, and using the continuity of cosine we get that l = cos l . l=\cos l. Solving this equation approximately we get that l 0.739. l\approx 0.739. And this is the answer of the question.

Awesome solution! +1

Prasun Biswas - 5 years, 8 months ago

Log in to reply

Thank you, Prasun!

Arturo Presa - 5 years, 8 months ago

However, by calculator, I get the limit to be 0.999847741....

A Former Brilliant Member - 4 years, 10 months ago

Log in to reply

That's the solution to cos ( l ) = l \cos(l^\circ)=l , i.e., when the angle in the cosine is in degrees. When we talk about the solution to cos l = l \cos l=l , the angle l l inside the cosine is considered to be in radians.

I guess you forgot to change the mode to "radians" in your calculator.

Prasun Biswas - 4 years, 10 months ago

What an excellent and perfect solution!

A Former Brilliant Member - 4 years, 10 months ago

Xn=??????why |Xn|≤1?

张 艺腾 - 3 years, 6 months ago

Log in to reply

Because the cosine of a real number is always bounded by -1 and 1

Andre Bourque - 2 years, 11 months ago

Assume that limit of the expression exist and it is x x . Then, we have:

x = lim n ( cos ( cos ( cos ( . . . ) ) ) ) = cos ( lim n ( cos ( cos ( cos ( . . . ) ) ) ) ) = cos x \begin{aligned} x & = \lim_{n \to \infty} {(\cos{(\cos{(\cos{(...)})})})} \\ & = \cos{\left( \lim_{n \to \infty} {(\cos{(\cos{(\cos{(...)})})})}\right)} \\ & = \cos{x} \end{aligned}

Using numerical method, we get x = 0.739 x=\boxed{0.739}

Moderator note:

1.You cannot simply assume that the limit exists.
2. f ( x ) = x f(x) =x is not a sufficient condition to guarantee that we will converge there.

For example, if f ( x ) = x 2 f(x) = x^2 , then we will not have lim f ( n ) ( x ) = 1 \lim f^{(n)}(x) = 1 unless we already started at x = 1 x = 1 .

For such limit of functions questions, more work will need to be done.

I actually didn't simply assume. I assumed because I didn't know. I was just saying if the limit exists the solution is as follows. Coming from a public relations background, I feel that it would sound better if we say "We shouldn't assume" than "You cannot simply assume". If it is something bad use "we". If it is something good use "you". It does the same job, but the reader feels better.

Okay, back to math.

My argument is as follows:

For any real number x x there exists a unique real value of cos x [ 1 , 1 ] \cos{x} \in [-1,1] . Therefore, there also exists a unique value of cos ( cos x ) \cos{(\cos{x})} , cos ( cos ( cos x ) ) \cos{(\cos{ (\cos{x})})} , cos ( cos ( cos ( cos x ) ) ) \cos{(\cos{(\cos{(\cos{x})})})} , ..., implying a unique limit of the expression exists [ 1 , 1 ] \in [-1,1] . we know that if there is a real value α [ 1 , 1 ] \alpha \in [-1,1] such that cos α = α \cos{\alpha} = \alpha , then α \alpha is the limit. This is because α = cos α \alpha = \cos{\alpha} = cos ( cos α ) = \cos{(\cos{\alpha})} = cos ( cos ( cos α ) ) = \cos{(\cos{(\cos{\alpha})})} = . . . =... . Since cos x \cos{x} is continuous [ 1 , 1 ] \in [-1,1] , and α [ 1 , 1 ] \alpha \in [-1,1] , there must exists α \alpha such that cos α = α \cos{\alpha} = \alpha .

Chew-Seong Cheong - 6 years ago

Log in to reply

Never made that observation about pronouns before. Great solution, as well.

Jake Lai - 6 years ago

Log in to reply

I don't think the "we" and "you" are relevant.

Pi Han Goh - 6 years ago

Sorry, Jake, are you the challenge master. Anyway, I read it as the challenge master is assuming or accusing me of simply assuming. It makes a difference. I have been getting paid for writing public-relations stuff. As an example. consider two girls meet after years. The prettier one tells her obviously not-so-pretty friend, "You are so pretty." Imagine how her friend will feel. No image, how her friend will feel, if she says, "We are so pretty."

Chew-Seong Cheong - 6 years ago

Log in to reply

@Chew-Seong Cheong Nope, that's Calvin Lin. Yeah, makes sense.

Jake Lai - 6 years ago

You only seem to show that if there is a real value α \alpha such that cos α = α \cos{\alpha} = \alpha , then α \alpha is the limit but, what about rest of the numbers?

Abhishek Sharma - 6 years ago

Log in to reply

It is shown that:

x = lim n ( cos ( cos ( cos ( . . . ) ) ) ) = cos ( lim n ( cos ( cos ( cos ( . . . ) ) ) ) ) = cos x \begin{aligned} x & = \lim_{n \to \infty} {(\cos{(\cos{(\cos{(...)})})})} \\ & = \cos{\left( \lim_{n \to \infty} {(\cos{(\cos{(\cos{(...)})})})}\right)} \\ & = \cos{x} \end{aligned}

As the Challenge Master has mentioned, this is true only if the series converges or the limit exists. So it is true, if the limit exists. Unless, we can get another x = f ( x ) x=f(x) , where f ( x ) cos x f(x)\ne \cos{x} . Because there is only one unique cos x \cos{x} for every x x in the domain.

Chew-Seong Cheong - 6 years ago

x R , cos x [ 1 , 1 ] cos ( cos x ) [ cos 1 , cos 0 ] = [ cos 1 , 1 ] \forall~x\in\Bbb R~,~\cos x\in [-1,1]\implies \cos(\cos x)\in [\cos 1,\cos 0]=[\cos 1, 1]

After the second cosine calculation, the domain for the rest of the composition becomes [ cos 1 , 1 ] [\cos 1,1] and subsequent cosine calculations will result in domains which are proper subsets of this interval. I think a sufficient reasoning here would be that the function f ( x ) = cos ( x ) f(x)=\cos(x) is a obviously a bijection in [ cos 1 , 1 ] [\cos 1,1] . Because of this, if a limit exists, it is guaranteed to be unique.

Showing that the limit exists requires some work to be done. One can note that the subsequent domains we get are proper subset of the previous domain we had. As the composition continues, the domain gets more and more restricted. A time will come when the domain converges to a single point which is the unique limit we need. Of course, this is just an informal explanation, not a rigorous proof.

Prasun Biswas - 5 years, 12 months ago

'Numerical method'? What's that? You(or We, as you like) must be meaning Newton Method, I guess.

Kartik Sharma - 6 years ago

Log in to reply

There are many numerical methods for root approximation. For example, we have Secant method , Muller's method and many more.

Of course, Newton's Method (better known as Newton-Raphson method) is significantly popular and easier to use.

Prasun Biswas - 5 years, 12 months ago

Log in to reply

I have heard of Secant Method but Muller's method reminds me of Thomas Mueller, oh yeah! BTW, Newton Method better known as Newton-Raphson or Newton-Raphlson ?

Kartik Sharma - 5 years, 12 months ago

Log in to reply

@Kartik Sharma Newton-Raphson because it was created by Newton and Joseph Raphson.

Prasun Biswas - 5 years, 12 months ago

Log in to reply

@Prasun Biswas Oh I thought it was Joseph Raphlson. Thanks for the clearing that doubt.

Kartik Sharma - 5 years, 12 months ago

Log in to reply

@Kartik Sharma Mention not. :)

Prasun Biswas - 5 years, 12 months ago

I tried to explain the convergence of the sequence in my solution that I also included here as a separate post.

Arturo Presa - 5 years, 8 months ago

@Chew-Seong Cheong I learned too much from your solutions it's simple and smart thanks alot

A Former Brilliant Member - 2 years, 10 months ago

Log in to reply

But the solution was not complete.

Chew-Seong Cheong - 2 years, 10 months ago

Log in to reply

You got the right answer we are not looking for proofs !

A Former Brilliant Member - 2 years, 10 months ago
Jayme Mendes
Jun 8, 2015

Assuming that x is such that

0<cosx<1,

then:

cos1<cos(cosx)<1

cos1<cos(cos(cosx))<cos(cos1))

cos(cos(cos1))<cos(cos(cos(cosx)))<cos(cos1))

and so on. These cosines vary less and less, alternating up and tending the limits at every step.It can be shown that after a certain number of cases we have, to three decimal places, the threshold value 0739.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...