Question:
Prove that if $\lim_{n \rightarrow \infty }\sum u_n=l\Rightarrow \lim_{n \rightarrow \infty} u_n=0$ by using Cauchy definition of convergence.
My answer:
1- First of all let's write: $S_n=\sum_{k=1}^{n} u_k$. As it is given by assumption we have that $\lim_{n \rightarrow \infty }S_n=l$ so $S_n$ converges and is a Cauchy sequence.
2- Now let's choose one $\epsilon_0>0$ and WLOG we choose $p>q$ and so $S_p-S_q=\sum_{k=q+1}^{p} u_k$.
By definition of a Cauchy sequence it exists at least one $N_{\epsilon_0} \in \mathbb{N} $ s.t. $\forall p,q>N_{\epsilon_0} \Rightarrow |S_p-S_q|<\epsilon_0$.
Now if it is true $\forall p,q>N_{\epsilon_0}$ it is in particular true for the couple $\left (p;q \right )$ choosed as follow $\left ( q, p=q+1 \right )$ and for the couple $\left ( q+1; p=q+2 \right )$...etc... In such cases we get $|S_p-S_q|=|u_p|<\epsilon_0$. In other words for the given $\epsilon_0>0 $, $\exists N_{\epsilon_0} \in \mathbb{N}$ s.t. we have an infinity element of the form $u_{q+k}$ in the interval $\left ( -\epsilon_0, \epsilon_0 \right )$ with $q>N_{\epsilon_0}, k \in \mathbb{N}$.
3- The same reasoning is valid $\forall \epsilon>0$ (trivial).
Hence it means that $\forall \epsilon>0$ we will always be able to find a $ N_{\epsilon} \in \mathbb{N}$ such that there will be an infinite amount element of the form $u_{q+k}, k \in \mathbb{N}$ in the interval $\left ( -\epsilon, \epsilon \right )$ at the condition that $q> N_{\epsilon}$. And this is exactly the definition of $\lim_{n \rightarrow \infty} u_n=0$.
Q.E.D.
Summarize, my prove is base on the fact that because $S_n$ is a Cauchy sequence, by choosing wisely $p,q>N$ of the definition we can see that the Cauchy definition defines too a sequence $u_n$ that converges to zero. I hope it is clear enough.
Is it correct?
In order to prevent the following confusion undersigned by @Andrew that my previous prove: " suggests that you do not really understand limits."[...] " that there is a subsequence which converges, not the entire sequence.".
I think I can be more precise and re-write my proof.
1- First of all let's write: $S_n=\sum_{k=1}^{n} u_k$. As it is given by assumption we have that $\lim_{n \rightarrow \infty }S_n=l$ so $S_n$ converges and is a Cauchy sequence.
2- Now let's choose one $\epsilon_0>0$ and WLOG we choose $p>q$ and so $S_p-S_q=\sum_{k=q+1}^{p} u_k$.
By definition of a Cauchy sequence it exists at least one $N_{\epsilon_0} \in \mathbb{N} $ s.t. $\forall p,q>N_{\epsilon_0} \Rightarrow |S_p-S_q|<\epsilon_0$. Now if it is true $\forall p,q>N_{\epsilon_0}$ it is in particular true for all the couples $\left (p;q \right )$ choosed as follow $\left ( q, p=q+1 \right )$ and for the couple $\left ( q+1; p=q+2 \right )$...etc... In such cases we get $|S_p-S_q|=|u_p|<\epsilon_0$. In other words for the given $\epsilon_0>0 $, $\exists q_{\epsilon_0} > N_{\epsilon_0} \in \mathbb{N}$ s.t. $\forall k \in \mathbb{N} \Rightarrow u_{q+k} \in \left( -\epsilon_0, \epsilon_0 \right )$.
In words it means that a direct consequence of the Cauchy definition is that we can wisely choose couples $\left( p;q \right )$ s.t. we will get that all the elements of the form/sequence $u_{q+k}$ will be in the interval $\left( -\epsilon_0, \epsilon_0 \right )$
3- The same reasoning is valid $\forall \epsilon>0$ (trivial). So we get exactly the definition of $\lim_{n \rightarrow \infty} u_n=0$.
Q.E.D.
Is it better? Correct?