2
$\begingroup$

I would like to have your help about the proof of the following statement:

If the sequences of complex numbers $\{F_N\}_{N \in \mathbb{N}},\{G_N\}_{N \in \mathbb{N}}$ have the following properties:

(i) $\lim_{N \rightarrow \infty} |F_N|^{1/N}=0;$

(ii) $\limsup_{N \rightarrow \infty} |G_N|^{1/N}\leq 1;$

(iii) there exists $N_0\in \mathbb{N}$ and $C>0$ such that $|F_N|\leq C|G_N| \sum_{k=N+1}^{\infty} |F_k|,$ for all $N \geq N_0$,

then there exists $N_1\in \mathbb{N}$ such that $F_N=0$ for all $N \geq N_1.$

This statement is in a paper without proof. Moreover, I only know the prove of this statement for the case when $G_N=1$ for all $N.$

The following is the proof for this special case.

Given the assumptions, there exists $M$ such that for all $ N\ge M,$ $$ |F_N|^{1/N}<\frac{1}{C+2}, \quad \quad \quad \quad |F_N|\le C\sum_{k=N+1}^\infty|F_k|.$$ We claim that for those $N$'s, $$|F_N|\le\left(\frac{C}{C+1}\right)^n\left(\frac{1}{C+2}\right)^N$$ for any non-negative integer $n$. Then, letting $n\rightarrow \infty,$ we see that $|F_N|=0.$

To prove the claim, we use induction on $n.$ When $n=0,$ the formula follows immediately from $|F_N|^{1/N}<1/(C+2).$ In general, using induction it follows that $$|F_N|\le C\sum_{k=N+1}^\infty|F_k|\le C\sum_{k=N+1}^{\infty}\left(\frac{C}{C+1}\right)^n\left(\frac{1}{C+2}\right)^k=\left(\frac{C}{C+1}\right)^{n+1}\left(\frac{1}{C+2}\right)^N.$$

Could you provide me a proof of the statement or give me an idea of the proof? Thank you very much.

Masik

$\endgroup$
2
  • 2
    $\begingroup$ How about $F_N = N^{-N}$ and $G_N = N$ as a counterexample? Here you have the first two conditions met and the inequality $N^{-N} \leq 2e N(N+1)^{-N-1}$ means $F_N \leq CG_N F_{N+1},$ which is even stronger than the third condition. Maybe the authors meant for $G_N$ to be bounded? $\endgroup$ Commented Oct 26, 2017 at 21:51
  • $\begingroup$ The paper states exactly what I posted on mathoverflow. I think that this paper contains an error. Thank you very much for your answer. Your counterexample is quite nice. $\endgroup$
    – Masik Kara
    Commented Oct 27, 2017 at 5:34

0

You must log in to answer this question.