3

Question: Show that for a sequence $\{x_m\}$ of real numbers to be a Cauchy sequence, it is necessary, but not sufficient that $|x_{m+1}-x_m|$ converges to zero.

This is how I proved that it is necessary: for $\{x_m\}$ to be cauchy, for all $\epsilon \gt 0$ there exists $n_o \in \mathbb{N}$ such that $|x_a-x_b| \lt \epsilon$ for all $a,b \ge n_o$. Now we can choose $a=m+1$ & $b=m$ such that both $a,b \ge n_o$. and by increasing or decreasing $m$ we can always ensure that for all $\epsilon$, $|x_{m+1}-x_m| \lt \epsilon $. Is this proof correct?

How to prove that it is not sufficient condition?

Mathy
  • 289
  • I know this question is very simple but even then please guide me if I am going in the right direction. – Mathy Jun 24 '13 at 02:14
  • 1
    See this post for examples showing the condition is not sufficient. – David Mitra Jun 24 '13 at 02:20
  • It means the way I have proved necessary condition is also wrong? – Mathy Jun 24 '13 at 02:23
  • No, your proof of necessary is right. As for not sufficient, you need to give an example of a sequence which satisfies the condition, but does not satisfy the condition of being Cauchy. David's link provides good examples. – Calvin Lin Jun 24 '13 at 02:27
  • The right idea seems to be there, but your phrasing could be better. Say: "Let $\epsilon>0$. Choose $N$ so that $|x_a-x_b|<\epsilon$ for all $a,b>N$. Then in particular, for $m>N$, $|x_{m+1}-x_m|<\epsilon$. So..." – David Mitra Jun 24 '13 at 02:27
  • Yeah I read those. They are awesome. Infact Nick is very right when he says "The "for all" quantifier on m and n is what confuses people about Cauchy sequences." – Mathy Jun 24 '13 at 02:28
  • thanx @DavidMitra – Mathy Jun 24 '13 at 02:31

0 Answers0