I'm working on a proof of the following theorem (from T.Tao's Analysis 2 book):

"Let $(X,d)$ be a metric space.

(a) Given any Cauchy sequence $(x_n)_{n=1}^\infty$ in $X$, we introduce the formal limit $LIM_{n\to\infty} x_n$. We say that two formal limits $LIM_{n\to\infty} x_n$ and $LIM_{n\to\infty} y_n$ are equal if $\lim_{n\to\infty} d(x_n,y_n)=0$. Show that this equality relation obeys the reflexive, symmetry, and transitive axioms. (DONE)

(b) Let $\overline{X}$ be the space of all formal limits of Cauchy sequences in $X$ with the above equality relation. Define a metric $d_{\overline{X}}\colon \overline{X}\times\overline{X}\to\mathbb{R}^+$ by setting $d_{\overline{X}}(LIM_{n\to\infty}x_n,LIM_{n\to\infty}y_n):=\lim_{n\to\infty} d(x_n,y_n).$ Show that this function is well-defined and gives $\overline{X}$ the structure of a metric space. (DONE)

(c) Show that the metric space $(\overline{X},d_{\overline{X}})$ is complete."

I've managed to prove the first two points but I'm having difficulties with the third one: I'm finding it hard to use the concept of Cauchy sequence of formal limits of Cauchy sequences to show that a Cauchy sequence in $\overline{X}$ has a limit in $\overline{X}$ so I'd appreciate any help in proving this point.

NOTE: I've already tried searching for similar questions but the ones I've found use the concepts of isometry and dense subset which are not discussed in the book I'm reading.


Let $(\bar{x}_k)_{k=1}^{\infty}$ be a Cauchy sequence in $\bar{X}$. For each $k$, we can represent $\bar{x}_k$ as a formal limit of a Cauchy sequence in $X$. In particular $\bar{x}_k = LIM_{n\rightarrow \infty}x_{k,n}$, where $(x_{k,n})_{n=1}^{\infty}$ is Cauchy in $X$.

Note: We don't want to take just any representative sequences. It makes the argument easier if we use sequences $(x_{k,n})_{n=1}^{\infty}$ satisfying $d(x_{k,n},x_{k,n+1})<2^{-n}$. This ensures that the sequences converge uniformly to their abstract limit points. We can always find such a sequence as a subsequence of any sequence converging to $\bar{x}_k$. Also, parts (a) and (b) above show that we are allowed to work with any particular representative that we like.

Now we write out the terms of the sequences \begin{align} \bar{x}_1 &\sim x_{1,1}, x_{1,2}, x_{1,3}, x_{1,4}, \dots\\ \bar{x}_2 &\sim x_{2,1}, x_{2,2}, x_{2,3}, x_{2,4}, \dots\\ \bar{x}_3 &\sim x_{3,1}, x_{3,2}, x_{3,3}, x_{3,4}, \dots\\ \bar{x}_4 &\sim x_{4,1}, x_{4,2}, x_{4,3}, x_{4,4}, \dots\\ \vdots & \hspace{1cm} \vdots \end{align} ans we claim that the diagonal sequence $(x_{n,n})_{n=1}^{\infty}$ is Cauchy in $X$, and $\bar{x}=LIM_{n\rightarrow \infty}x_{n,n}$ is the limit point of $(\bar{x}_k)_{k=1}^{\infty}$.

Step 1: $(x_{n,n})_{n=1}^{\infty}$ is Cauchy: The sequence $(\bar{x}_k)_{k=1}^{\infty}$ being Cauchy in $\bar{X}$ means that for every $\epsilon>0$, there is an $N\in \mathbb{N}$ such that $d_{\bar{X}}(\bar{x}_k,\bar{x}_n)<\epsilon$ when $k,n>N$, which means \begin{equation} \lim_{m\rightarrow\infty}d(x_{k,m},x_{n,m})<\epsilon. \end{equation}

So for $k,m,n>N$, we have \begin{align} d(x_{k,k},x_{n,n}) &\leq d(x_{k,k},x_{k,m})+d(x_{k,m},x_{n,m})+d(x_{n,m},x_{n,n})\\ &\leq 2^{-N}+d(x_{k,m},x_{n,m})+2^{-N} \end{align} Taking the limit as $m\rightarrow\infty$, we find $d(x_{k,k},x_{n,n})\leq 2^{-N+1}+\epsilon$, and we can make this as small as we want by increasing $N$.

Step 2: $\bar{x}$ is the limit of $(\bar{x}_k)_{k=1}^{\infty}$:

For $k>N$, we have \begin{align} d_{\bar{X}}(\bar{x},\bar{x}_{k}) &\leq \sup_{m>N} d(x_{m,m},x_{k,m})\\ &\leq \sup_{m>N} d(x_{m,m},x_{k,k})+d(x_{k,k},x_{k,m})\\ &\leq (2^{-N+1}+\epsilon) + 2^{-N}\\ &\leq 2^{-N+2}+\epsilon \end{align}

  • $\begingroup$ excellent answer, thank you. I will award the bounty to your answer asap. $\endgroup$ – lorenzo Jul 12 '17 at 19:27

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.