Let $(a_n)_{n\in\mathbb N}$ and $(b_n)_{n\in\mathbb N}$ be two real sequences such that

  • $a_n\geq0$ and $b_n\geq0$ for all $n\in\mathbb N$;
  • $\sum_{n=1}^{\infty} a_n=\infty$;
  • $(b_n)_{n\in\mathbb N}$ is non-increasing ($b_1\geq b_2\geq\ldots$);
  • $\lim_{n\to\infty} b_n=0$; and
  • $\sum_{n=1}^{\infty}a_nb_n<\infty$.

I want to show that $\sum_{n=1}^Na_n=o(b_N^{-1})$, that is, $$\lim_{N\to\infty}\left\{b_N\sum_{n=1}^N a_n\right\}=0.$$ In words, the sequence $(b_n)_{n\in\mathbb N}$ converges so quickly that it not only soothes but suppresses the divergence of $\sum_{n=1}^Na_n$.

Any hints would be much appreciated.

My work: it is fairly clear that the sequence in question is bounded, that is, $\sum_{n=1}^N a_n=O(b_N^{-1})$, as for any $N\in\mathbb N$, one has that $$b_N\sum_{n=1}^N a_n=\sum_{n=1}^N a_nb_N\leq\sum_{n=1}^Na_n b_n\leq\sum_{n=1}^{\infty}a_nb_n(<\infty),$$ but I can’t seem to prove convergence to $0$.

  • 1
    $\begingroup$ This is reminding me of summation by parts ... $\endgroup$ – Michael Burr May 1 '16 at 22:39

I got it. Any feedback on my solution would be appreciated.

Fix $\varepsilon>0$. Since $\sum_{n=1}^{\infty}a_n b_n<\infty$, there exists some $N_0\in\mathbb N$ such that $$\sum_{n=N_0+1}^{\infty}a_n b_n<\frac{\varepsilon}{2}.$$ Since $\lim_{n\to\infty} b_n=0$, there is some $N_1\in\mathbb N$ such that if $N\in\mathbb N$ and $N\geq N_1$, then $$b_N<\frac{\varepsilon}{2\left(\sum_{n=1}^{N_0} a_n+1\right)}$$ (the $+1$ added to deal with a potentially vanishing denominator). Hence, if $N\in\mathbb N$ and $N\geq \max\{N_0+1,N_1\}$, then one has that $$0\leq b_N\sum_{n=1}^Na_n= b_N\sum_{n=1}^{N_0}a_n+\sum_{n=N_0+1}^{N}a_n b_N<\frac{\varepsilon}{2}+\sum_{n=N_0+1}^{N}a_n b_n<\frac{\varepsilon}{2}+\frac{\varepsilon}{2}=\varepsilon.$$ (Note the proof goes through the same way even if $\sum_{n=1}^{\infty} a_n<\infty$.)

  • 1
    $\begingroup$ I think you got it; looks perfect. $\endgroup$ – Fimpellizieri May 1 '16 at 23:48
  • $\begingroup$ @Fimpellizieri The solution looked too smooth as compared to the time it took me to figure it out, so I was a little skeptical about its validity—that’s why I appreciate the feedback. Thank you very much. $\endgroup$ – triple_sec May 1 '16 at 23:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.