3

Let's take $\alpha$ and $\beta$ two reals in $(0, \frac{\pi}{2})$. Let's take $\lambda \in (0,1)$. Let's define the sequence $(u_n)$ as follows: \begin{eqnarray} u_n & = & \frac{\cos(\alpha - n \beta) - \lambda \cos(\alpha + n \beta)}{ \cos^n(\beta) } \end{eqnarray}

  1. Can we extract a subsequence $(u_{\phi(n)})$ that converges to a real number ? ( where $\phi$ is a strictly increasing application from $\mathbb{N} \to \mathbb{N}$ )

  2. If not, can we prove that the limit of $|u_n|$ is $+\infty$?

My intuition: I think It is possible to construct such a subsequence since we can always construct $\phi$ such that $\cos(\alpha - \phi(n) \beta) - \lambda \cos(\alpha + \phi(n) \beta)$ converges to $0$. The problem is that I am struggling with is how then to construct $(u_{\phi(n)})$ such that it converges.

Also, I had the idea is to use the Tschebyscheff polynomials (Chebyshev Polynomials)(https://en.wikipedia.org/wiki/Chebyshev_polynomials) as we can always express $u_n$ in function of chebyshev polynomials. I don't know yet how to move it forward so far. So if anyone has any idea, please share it here. Thank you in advance.

  • 1
    Welcome to MSE. Your question is phrased as an isolated problem, without any further information or context. This does not match many users' quality standards, so it may attract downvotes, or be closed. To prevent that, please [edit] the question. This will help you recognize and resolve the issues. Concretely: please provide context, and include your work and thoughts on the problem. These changes can help in formulating more appropriate answers. – José Carlos Santos Jul 06 '22 at 06:23
  • I presume that () means open interval? In any case, do you want a proof/algoirthm for-all(variables) or for-what(variables)? For instance, for an answer: can beta be specified rational, or must the proof include beta transcendental? – rrogers Jul 12 '22 at 20:09
  • Hi rrogers, Yes $(a, b)$ means an open interval that excludes $a$ and $b$. Feel free to share all ideas. Be it for a specified $\beta$ or a general case. – InfiniteMath Jul 14 '22 at 03:32

1 Answers1

1

Here is an idea of solution that I would like your comment on. Let's first write the following: \begin{eqnarray} \hspace{0.0cm} A & = & \cos (\alpha) (1-\lambda) \\ B & = & \sin (\alpha) (1 + \lambda) \\ \cos (\alpha - \beta n ) - \lambda \, \cos (\alpha + \beta n ) & = & A \cos (\beta n) + B \sin ( \beta n ) \hspace{1.cm} \\ & = & \sqrt{A^2 + B^2} \, \sin ( a_0 + \beta n ) \end{eqnarray} Where $a_0 \in (0, \frac{\pi}{2})$ ( since $\alpha \in (0, \frac{\pi}{2})$ ) . \begin{eqnarray} \hspace{0.0cm} \sin ( a_0 ) & = & \frac{ A }{ \sqrt{A^2 + B^2} } \\ \cos ( a_0 ) & = & \frac{ B }{ \sqrt{A^2 + B^2} } \end{eqnarray}

And therefore \begin{eqnarray} \hspace{0.0cm} u_n & = & \frac{ \sqrt{A^2 + B^2} \, \sin ( a_0 + \beta n ) } { \cos^n (\beta ) } \hspace{1.cm} \end{eqnarray}

We have two cases:
Case 1: $\beta = \frac{\pi p - a_0 }{q}$ and $a_0 = \frac{r \pi}{s}$ where $(r,s) = 1$, $p, q, r, s \in \mathbb{N}$.

In this case we have for each $n \in \mathbb{N}$: \begin{eqnarray} \sin ( a_0 + q (n s + 1) \beta ) & = & (-1)^{(n s + 1)p + 1} \sin ( \pi n r ) = 0 \hspace{1.0cm} \end{eqnarray} And therefore for each $n \in \mathbb{N}$: \begin{eqnarray} u_{ q (n s + 1) } & = & 0 \hspace{1.cm} \end{eqnarray} And hence a subsequence of $(u_n)$ that converges to zero.

Case 2: $\beta = \lim_{m \to +\infty } \frac{s_m p_m - r_m }{s_m q_m} \pi $ and $a_0 = \lim_{m \to +\infty } \frac{r_m }{s_m} \pi $ where $(r_m,s_m) = 1$, $p_m, q_m, r_m, s_m \in \mathbb{N}$.

Thanks to the density of $\mathbb{Q}$ in $\mathbb{R}$, we can find the sequences $(\frac{\pi p_m - a_0 }{q_m})$ and $(\frac{\pi r_m }{s_m})$ that converges respectively to $\beta$ and $a_0$.

Without loss of generality, we can assume that the sequence $(q_m)$ is increasing.

Intuitively, if it works in Case 1, it should work in Case 2 thanks to the density of $\mathbb{Q}$ in $\mathbb{R}$.

For each $n$, we can write that: \begin{eqnarray} \sin ( a_0 + n \beta ) & = & \lim_{m \to +\infty } \sin ( \frac{r_m }{s_m} \pi + n \frac{s_m p_m - r_m }{s_m q_m} \pi ) \hspace{1.0cm} \\ u_n & = & \lim_{m \to +\infty } \frac{ \sqrt{A^2 + B^2} \, \sin ( \frac{r_m }{s_m} \pi + n \frac{s_m p_m - r_m }{s_m q_m} \pi ) } { \cos^n (\beta ) } \hspace{1.cm} \end{eqnarray} And \begin{eqnarray} \hspace{-1.cm} \lim_{n \to +\infty } u_n & = & \lim_{n \to +\infty } \lim_{m \to +\infty } \frac{ \sqrt{A^2 + B^2} \, \sin ( \frac{r_m }{s_m} \pi + n \frac{s_m p_m - r_m }{s_m q_m} \pi ) } { \cos^n (\beta ) } \hspace{1.cm} \end{eqnarray} If we take $n$ of the form $n_m = q_m ( 1 + m \prod^{m}_{i=1} s_i )$, we will have (subject to justification!): \begin{eqnarray} \hspace{-1.cm} \lim_{m \to +\infty } \sin ( a_0 + n_m \beta ) & = & \lim_{m \to +\infty } \underbrace{ \sin ( \frac{r_m }{s_m} \pi + q_m ( 1 + m \prod^{m}_{i=1} s_i ) \frac{s_m p_m - r_m }{s_m q_m} \pi ) }_{ = 0} \hspace{1.0cm} \\ & = & \lim_{m \to +\infty } 0 = 0 \end{eqnarray} And \begin{eqnarray} \hspace{-1.12cm} \lim_{m \to +\infty } u_{ n_m } & = & \lim_{m \to +\infty } \underbrace{ \frac{ \sqrt{A^2 + B^2} \, \sin ( \frac{r_m }{s_m} \pi + q_m ( 1 + m \prod^{m}_{i=1} s_i ) \frac{s_m p_m - r_m }{s_m q_m} \pi ) } { \cos^{n_m} (\beta ) } }_{ = 0} \hspace{1.cm} \\ & = & \lim_{m \to +\infty } 0 = 0 \end{eqnarray} Note that the sequence $(n_m)$ is strictly increasing since the sequence $(q_m)$ is increasing and the sequence $ ( 1 + m \prod^{m}_{i=1} s_i )$ is strictly increasing. And hence a subsequence of $(u_n)$ that converges to zero.

Please let me know if there is any problem with such solution. Thank you.

  • Hmm, I think my main problem with this is that you didn't use the fact that the denominator is decreasing exponentially anywhere in your proof. You have shown that there exists a subsequence such that the numerator goes to $0$, but this does not convince me that the 1 over the denominator cannot grow faster overall causing the entire expression to go to infinity. – QC_QAOA Jul 18 '22 at 12:35