4
$\begingroup$

Let $\left\{a_k\right\}_{0\leqslant k\leqslant n}$ and $\left\{b_k\right\}_{0\leqslant k\leqslant n}$ be two positive decreasing sequences such that $c_1a_k\leqslant b_k\leqslant c_2 a_k$ for all $k$ for some positive constants $c_1,c_2$. Let $\epsilon>0$ and $f>0$ a function such that $$\sum_{k=0}^na_k\sin(k\epsilon)\geqslant f(n).$$

Question. Is it true that $$\sum_{k=0}^nb_k\sin(k\epsilon)\geqslant cf(n)$$ for some constant $c$ (independent of $n$, $\epsilon$)?


Without the monotonicity condition on the $a_k$'s and $b_k$'s, the answer is no, as shown in this answer.

$\endgroup$
14
  • $\begingroup$ Are you assuming that $f>0$? $\endgroup$ Commented Apr 21, 2021 at 15:17
  • $\begingroup$ Yes I am_______ $\endgroup$
    – coco
    Commented Apr 21, 2021 at 15:19
  • 3
    $\begingroup$ Probably because there was 40 minutes between your two question, leaving zero time for you to try to figure it out yourself. $\endgroup$
    – username
    Commented Apr 21, 2021 at 15:40
  • 1
    $\begingroup$ @coco : I have edited your post to make the connection with the previous post explicit. $\endgroup$ Commented Apr 21, 2021 at 16:07
  • 1
    $\begingroup$ I upvoted, since the downvotes are unjustified (imo) $\endgroup$ Commented Apr 21, 2021 at 18:06

1 Answer 1

4
$\begingroup$

$\newcommand\ep\varepsilon$The answer is still no (assuming you wanted $c>0$). Indeed, let $$s_n:=\sum_{k=0}^n a_k\sin(k\ep),\quad t_n:=\sum_{k=0}^n b_k\sin(k\ep),$$ where $\ep=3\pi/4$, $a_0=b_0=10$, $a_1=1$, $b_1=9$, and $a_k=b_k=2^{-k}$ for $k\ge2$.

Then $c_1a_k\le b_k\le c_2a_k$ for $c_1=1$, $c_2=9$, and all $k\ge0$. Moreover, $s_0>0$ and $s_1=-1 + 5\sqrt2>6>0$, whence $s_n>6-\sum_{k=2}^\infty 2^{-k}>0$ for $n\ge2$. So, $s_n>0$ for all $n\ge0$. So, letting $f(n):=s_n$ for all $n\ge0$, we get a positive function $f$ such that $s_n\le f(n)$ for all $n\ge0$. Thus, all your conditions on the $a_k$'s and $b_k$'s are satisfied.

However, $t_1=-9 + 5\sqrt2<-19/20<0$ and, moreover, $t_n<-19/20+\sum_{k=2}^\infty 2^{-k}<0$ for all $n\ge2$. Thus, the condition $t_n\le cf(n)$ fails to hold for any $c>0$ and any $n\ge1$.

$\endgroup$
7
  • $\begingroup$ Thank you very much Iosif. This is not a good news for my calculation, I should find more hypothesis to have this lower bound. $\endgroup$
    – coco
    Commented Apr 22, 2021 at 8:58
  • $\begingroup$ @coco : I understand that you hoped to get positive answers to your questions. However, the answers, even though negative, seem correct and complete, and they may help you find the right conditions. So, to have a closure, are you satisfied with the answers in this sense? $\endgroup$ Commented Apr 22, 2021 at 11:41
  • $\begingroup$ Yes, thank you for taking time to find a counter example. $\endgroup$
    – coco
    Commented Apr 22, 2021 at 12:13
  • $\begingroup$ @coco : Why have you unaccepted this answer? Is it not a correct and complete answer to your post? $\endgroup$ Commented Nov 21, 2023 at 16:31
  • $\begingroup$ waw i'm impressed by the rapidity of your reaction. No it was not at all answering my question, i said ok to make you happy because this was just a tiny part of my research and i could not put the whole context to explain what are my a_k and b_k and \epsilon, etc. But no, my \epsilon is not 3\pi/4 and your example of a_k and b_k only differ in a_1 and b_1 which is also not my case. But of course i was not going to write a new post "Oscillating sums III" adding these facts, because you don't get that this was not just an exercise but part of my research :) $\endgroup$
    – coco
    Commented Nov 22, 2023 at 8:32

You must log in to answer this question.

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