I want to check if my intuition about continuity is correct.

Suppose $(X, d_X)$ and $(Y, d_Y)$ are two metric spaces that are isometrically isomorphic, i.e., there is an isomorphism $h : X \to Y$ with $d_X(x_1,x_2) = d_Y(h(x_1), h(x_2))$ for all $x_1,x_2 \in X$. Let $M \subseteq X \cap Y$ be a common subset of $X$ and $Y$.

Is it true that if a function $f : M \to M$ is continuous in $X$, it is continuous also in $Y$, and vice versa?

In other word: Suppose we have shown, using the $\epsilon$-$\delta$ criterion, that $$ d_X(x,y) < \delta \quad\Longrightarrow\quad d_X(f(x), f(y)) < \epsilon. $$ Does this imply that $$ d_Y(x,y) < \delta' \quad\Longrightarrow\quad d_Y(f(x), f(y)) < \epsilon', $$ with possibly different $\delta'$ and $\epsilon'$?


As a more concrete example, consider the function $f : \mathbb{R}_{>0} \to \mathbb{R}_{>0}, \, x \mapsto \frac{1}{\sqrt{x}}$. Then $f$ is continuous in $(\mathbb{R}_{>0}, d_A)$ with distance function $d_A(x,y) = | \ln(x) - \ln(y) |$. Hint: Choose $\delta = \epsilon$ and show that $$ \begin{array}{ccccccccccc} \epsilon &=& \delta &>& d_A(x,y) &>& \left| -\dfrac{1}{2} \right| \cdot d_A(x,y) &=& \ldots &=& d_A(f(x), f(y)). \end{array} $$ Now notice that $(\mathbb{R}_{>0}, d_A)$ is isometrically isomorphic to $(\mathbb{R}, d_B)$ with $d_B(x,y) = | x - y |$. Is this fact already sufficient to conclude that $f$ is continuous also in $\mathbb{R}$?


You would also need that there is an isometric isomorphism from $X$ to $Y$ that fixes $M$ (equivalently, $d_X,d_Y$ agree on $M$). Otherwise, the set intersection of $X$ and $Y$ is not very meaningful, since it's changed by a simple relabelling of elements. For instance, for $X=\mathbb R$, you could replace each negative number $x$ with $(0,x)$, and apply some horrible bijection $[0,\infty)$ to relabel the nonnegative numbers, obtaining a metric space $Y$ which is isometrically isomorphic to $\mathbb R$. Then, $X \cap Y = [0,\infty)$, but your statement does not necessarily hold if our horrible bijection was not a homeomorphism; it is mere coincidence that the labelling of these elements coincide in $X$ and $Y$, and they needn't have the same metric space structure.

So if you assume that $d_X,d_Y$ agree on $M$, the statement is trivially true, since then $(M,d_X) = (M,d_Y)$.


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.