2

The number of real solutions of the equation $$\cos(\cos(\cos(\cos(x)))) = \sin(\sin(\sin(\sin(x))))$$

(I have no clue to solve this problem)

Axion004
  • 10,056
Anonymous
  • 400
  • 1
    @AryamanMaithani can please elaborate your working (i am a jee aspirant and this question came in in our advanced test) the given answer claims that $\cos(\cos(\cos(\cos(x)))) = \sin(\sin(\sin(\sin(x))))$ has no real solution, – Anonymous Jul 12 '20 at 19:10
  • 1
    I think i has no solution, as the maximum value of argument keeps reducing, I'm not very sure about it. But usually these type of problems just look ugly. – UmbQbify Jul 12 '20 at 19:18
  • can you please elaborate it @UmbQbify-Key20- – Anonymous Jul 12 '20 at 19:19
  • 1
    Strategy: by the functions' symmetries it suffices to consider $x\in[0,\frac\pi2]$. Both iterated functions are increasing on that interval. Estimate them both at say $x=0,\frac\pi4,\frac\pi2$. – Greg Martin Jul 12 '20 at 19:43
  • 2
    A nice answer for a duplicate is here: https://math.stackexchange.com/a/681395/752069 – VIVID Jul 12 '20 at 20:05

1 Answers1

5

\begin{align*} \cos(x) &\in [-1,1] \text{,} \\ \cos\cos(x) &\in [\cos(1),1] \text{,} \\ \cos\cos\cos(x) &\in [\cos\cos(1),\cos(1)] \text{, and} \\ \cos\cos\cos\cos(x) &\in [\cos\cos\cos(1),\cos\cos(1)] \text{.} \end{align*}

The second line is because cosine is even and attains its maximum, $1$ on the interval $[-1,1]$. We know $[\cos(1),1]$ only contains positive numbers (because $\cos(0) > 0$ and the least positive angle producing zero from the cosine is $\pi/2 > 1$).

\begin{align*} \sin(x) &\in [-1,1] \text{,} \\ \sin\sin(x) &\in [-\sin(1),\sin(1)] \text{,} \\ \sin\sin\sin(x) &\in [-\sin\sin(1),\sin\sin(1)] \text{, and } \\ \sin\sin\sin\sin(x) &\in [-\sin\sin\sin(1),\sin\sin\sin(1)] \text{.} \end{align*}

This last interval contains positive and negative numbers.

The only way there can be solutions is if the upper bound of the last interval for iterated sines is at least the lower bound of the last interval for iterated cosines. So, we must decide $$ \sin\sin\sin(1) \overset{?}{\geq} \cos\cos\cos(1) $$

One way to resolve this is with a calculator. \begin{align*} \sin\sin\sin(1) &= 0.6784304773607402289{\dots} \text{ and } \\ \cos\cos\cos(1) &= 0.6542897904977791499{\dots} \text{.} \end{align*} So the potential remains that there may be solutions. (And as observed in comments, by periodicity, there are either zero solutions or infinitely many.)

If $\sin(y)$ is going to be as large as the lower bound on the iterated cosines, it is at least as large as $1/2$. (This may seem like pulling a mysterious constant out of thin air; it is not. There are three angles in the first quadrant we are responsible for knowing very well. Their sines are $1/2$, $\sqrt{2}/2$, and $\sqrt{3}/2$. The other two are too large.) This requires $\pi/6 < y < 5\pi/6$ (up to an integer multiple of $2\pi$).

If $y = \sin(w)$, since $\pi/6 > 1/2$, the same argument requires $\pi/6 < w < 5\pi/6$ (up to an integer multiple of $2\pi$). (Now we have $\sin\sin(w) > 1/2$.)

If $w = \sin(z)$, since $\pi/6 > 1/2$, the same argument requires $\pi/6 < z < 5\pi/6$ (up to an integer multiple of $2\pi$). (Now we have $\sin\sin\sin(z) > 1/2$.)

Repeating the argument again, we find that for $\sin\sin\sin\sin(x) > 1/2$ it is the case that $\pi/6 < x < 5\pi/6$ (up to an integer multiple of $2\pi$). So any solution on $[0,2\pi]$ has $x \in [\pi/6, 5\pi/6]$.

For $x \in [\pi/6, 5\pi/6]$, \begin{align*} \cos(x) &\in [-1/2,1/2] \text{,} \\ &\vdots \\ \cos\cos\cos\cos(x) &\in [\cos\cos\cos(1/2),\cos\cos(1)] \\ &= [0.698{\dots}, 0.857{\dots}] \text{.} \end{align*}

So for the angles where the iterated sine is large enough to meet the range of the iterated cosine, the iterated cosine is greater than the maximum of the iterated sine. Therefore, there is no solution.

We can also see this from a plot of the two iterated functions.

Mathematica graphics

The upper bound of the orange trace is higher than the lower bound of the blue trace, but whenever the orange trace is $\geq 1/2$, the height of the blue trace is at least $0.698{\dots}$, which is above the upper bound of the orange trace ($0.678{\dots}$).

Eric Towers
  • 67,037
  • 1
    Suggestion: state the answer right up front, and then tell the story. – uniquesolution Jul 12 '20 at 19:51
  • 11
    @uniquesolution : I do not write jokes, stories, or mathematics backwards. To do so wastes the reader's time. – Eric Towers Jul 12 '20 at 19:52
  • 1
    You wrote a long argument, taking the reader through lengthy calculations, and you haven't stated the answer to the question: what is the number of solutions to the equation? You may as well just posted the graph. Time wasting? Think again. – uniquesolution Jul 12 '20 at 19:54
  • I think it is your prerogative to give the bottom line up front or not, but I point out that ostensibly, it is not giving it up front that wastes more time than giving it up. (Again, I'm not advocating any course of action.) – Brian Tung Jul 12 '20 at 19:55
  • 2
    @BrianTung : If I were writing a memorandum, which implicitly has the assumption that understanding by readers is not required, then a preliminary summary would be justified. Since mathematics is written under the assumption that understanding is required, a preliminary summary is incorrect. – Eric Towers Jul 12 '20 at 19:57
  • @uniquesolution : Although I agree that I omitted a clear statement of the conclusion. Thanks for the note! Fixed. – Eric Towers Jul 12 '20 at 19:59
  • OK, though "incorrect" sounds like you're prescribing that course of action for others. I'll choose for myself, thanks! :-) – Brian Tung Jul 12 '20 at 20:34
  • Btw, I tried to find complex roots for curiosity numerically, but no luck, except for guessed real part periodicity like $n \pi i$ – Narasimham Jul 12 '20 at 20:52
  • 1
    Very nice answer. – Sebastiano Jul 12 '20 at 21:23
  • 2
    Please search for dupes before answering a dupe. – amWhy Jul 13 '20 at 03:34
  • $\cos(1)\leq\cos\cos\cos(x)\leq\cos\cos(1).$ – Riemann Sep 05 '22 at 14:46