0
$\begingroup$

Assume I have a sequence $\{a_m\}$ that is vanishing and strictly positive: $$ 0<a_{m+1}\leq a_m\leq\ldots\leq a_1<\infty, \quad \lim_{m\to \infty}a_m = 0 $$ Is it true or false that this has a subsequence $\{a_{m_n}\}$ such that $$ a_{m_n}\asymp n^{-1} $$ which is to say, there exists a constant $C>0$ such that for all $n$: $$ \frac{1}{Cn} \leq a_{m_n} \leq \frac{C}{n} $$

$\endgroup$
6
  • $\begingroup$ Isn't this false for any sequence that goes to zero "quickly", e.g., $1/n^2$, or $1/2^n$? $\endgroup$ Commented Jun 9, 2020 at 3:20
  • $\begingroup$ No, if $a_m=1/m^2$, take $m(n) = \lceil \sqrt{n}\rceil$ $\endgroup$ Commented Jun 9, 2020 at 3:36
  • $\begingroup$ What would the subsequence be for $1/2^n$? $\endgroup$ Commented Jun 9, 2020 at 4:44
  • $\begingroup$ So you're allowing, say, $m(10)=m(11)=m(12)=\cdots=m(16)$? A subsequence can contain repeated values, even if those values aren't repeated in the source sequence? $\endgroup$ Commented Jun 9, 2020 at 5:46
  • $\begingroup$ Yes. The subsequence can have repeated values. $\endgroup$ Commented Jun 9, 2020 at 12:01

1 Answer 1

0
$\begingroup$

The answer is no. E.g., let $a_m:=1/(m!)^2$ and $n_k:=k!(k+1)!$ for natural $m,k$. Your desired condition means that $|\ln na_{m_n}|$ is bounded. In our case, we have $$a_{k+1}<\frac1{n_k}<a_k$$ for all $k$. So, for all $m\le k$ we have $1<n_k a_k\le n_k a_m$ and hence $|\ln(n_k a_m)|\ge|\ln(n_k a_k)|=\ln(k+1)$. Similarly, for all $m\ge k+1$ we have $1>n_k a_{k+1}\ge n_k a_m$ and hence $|\ln(n_k a_m)|\ge|\ln(n_k a_{k+1})|=\ln(k+1)$. So, $|\ln(n_k a_m)|\ge\ln(k+1)$ for all $k,m$.

So, for any $(m_n)$ $$|\ln n_ka_{m_{n_k}}|\ge\inf_m|\ln n_k a_m|\ge\ln(k+1)\to\infty$$ as $k\to\infty$, so that $|\ln na_{m_n}|$ is unbounded, for any choice of $(m_n)$.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged .