The question says this:

If $(X,d)$ is a metric space and $\{x_n\}$ and $\{y_n\}$ are Cauchy sequences, prove that $\{d(x_n,y_n)\}$ is a Cauchy sequence in $R$.

I see that I would have to show that $d_1(d(x_n,y_n), d(x_m,y_m)) < \epsilon $ for all $m,n \ge N $ for some metric on R. If I assume that $d_1$ is the Euclidean metric then the reverse triangle inequality leads to the sequence being Cauchy, but how would I show this for an arbitrary metric $d_1$ in R?


You need to do this for the standard metric on the reals. If not, it should have been specified, as Cauchy-ness depends on the metric used, and it can fail for other choices of the metric on the reals.

So you need to show that for every $\varepsilon > 0$ there exists some $N$ such that for all $n,m \ge N$ we have

$$\left| d(x_n, y_n) - d(x_m,y_m) \right| < \varepsilon\text{.}$$

So find $N_1$ for $(x_n)$ and $\frac{\varepsilon}{2}$, $N_2$ for $(y_n)$ and $\frac{\varepsilon}{2}$. Then

$$d(x_n, y_n) \le d(x_n, x_m) + d(x_m,y_m) + d(y_m, x_m)$$

so for $n,m \ge \max(N_1,N_2)$ we have

$$d(x_n, y_n) < \frac{\varepsilon}{2} + d(x_m,y_m) + \frac{\varepsilon}{2}$$

from which we have

$$ d(x_n, y_n) - d(x_m,y_m) < \varepsilon$$

and then we interchange $n,m$ in the above argument to get the other order on the left, and take the max to get the absolute value.

| cite | improve this answer | |

The intended metric in $\mathbb{R}$ is probably the usual metric here, namely the one defined by the absolute value of the difference. The proof is particularly simple if you are allowed to use a metric completion $\mathbf{X}$ of $X$. Passing to the limits $x=\lim_{n\to\infty}x_n$ and $y=\lim_{n\to\infty}y_n$ you can easily show from the triangle inequality that $d(x_n,y_n)$ necessarily converges to $d(x,y)$.

The existence of metric completion is immediate by passing to the ultrapower and does not require delicate analysis of Cauchy sequences.

| cite | improve this answer | |
  • $\begingroup$ This proof is part of the proof for the metric completion, I think. So that would be circular. $\endgroup$ – Henno Brandsma May 22 '16 at 14:59

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.