5
$\begingroup$

Please consider the following definitions:

A: Suppose $d$ and $e$ are metrics on a set $X$. Then, $d$ and $e$ are topologically equivalent metrics if and only if the identity functions from $(X,d)$ to $(X,e)$ and from $(X,e)$ to $(X,d)$ are both continuous.

B: Suppose $(X,d)$ and $(Y,e)$ are metric spaces. Then $X$ and $Y$ are said to be homeomorphic or topologically equivalent if and only if, there exists a bijective function $f:X \rightarrow Y$ that is continuous and has continuous inverse; such a function is called a homeomorphism.

C: Suppose $d$ and $e$ are metrics on a set $X$. If $d$ and $e$ are topologically equivalent, then certainly $(X,d)$ and $(X,e)$ are homeomorphic, because the identity function $I_{d,e}:(X,d) ]\rightarrow (X,e)$ is continuous and has a continuous inverse $I_{e,d}:(X,e) \rightarrow (X,d)$. However, the converse need not be true

In definition C, it is specifically mentioned that the converse need not be true. I assume, by the word : converse, the author implies : if $(X,d)$ and $(X,e)$ are homeomorphic, then, $d$ and $e$ are topologically equivalent.

I reason this as follows: there can exist bijective function $f:X \rightarrow Y$ that is continuous and has continuous inverse; but which is not the identity function. Hence, $d$ and $e$ need not be topologically equivalent even though $(X,d)$ and $(X,e)$ are homeomorphic.

Is this correct?

Thanks

$\endgroup$
5
$\begingroup$

Yes, this is correct. The "topological equivalence" of metrics only says that they induce the exact same topology.

Here's an example of a scenario the authors thinks about. Let $(X, d)$ and $(X, e)$ be two topologically not-equivalent metrics on a single space $X$ -- for example, let $X = \mathbb{R}^2$, $d$ is standard metric, and $e$ be the star metric: $e(x, y) = d(x, y)$ if $x$ and $y$ lie on the same straight line through origin, and $d(x, 0) + d(0, y)$ if they don't.

Consider now set $Y = X \coprod X$, that is, the disjoint union of two copies of $X$. Each element of $X$ appears twice in $Y$, we denote these two copies by $x^1$ and $x^2$. We can put two different metrics on $Y$ -- we can have a metric $d_1$ that is equal to $d$ on first copy, and to $e$ on the second copy, and always gives distance $1$ for points in two different copies. We can also have metric $d_2$ that is $e$ on first copy, and $d$ on second copy.

Then, the spaces $(Y, d_1)$ and $(Y, d_2)$ are homeomorphic (actually isometric), homeomorphism just swaps points in copies. However, the identity functions is not homeomorphism.

$\endgroup$
3
$\begingroup$

Yes, your statement of the converse is correct. To construct a counterexample, you might start with a space that admits an involution (order two symmetry), such as the plane and the involution being reflection in the $x$-axis. Put inequivalent metrics on the two halves such that they agree on the common overlap; call the resulting metric $d$. Then switch the two; call the resulting metric $e$, i.e. $e(x,y) = d(i(x), i(y))$, where $i:X \to X$ is the involution. The identity will not be a homeomorphism, but the involution will. I can provide two such metrics on the two half-planes if you like, or you can think about how to construct one. Or maybe there is an easier example.

$\endgroup$
  • $\begingroup$ Thanks. I would appreciate if you could provide me with the example you were referring to. $\endgroup$ – MathMan Apr 12 '18 at 1:27
  • 1
    $\begingroup$ It's very similar to xyzzyz's example: on the closed upper half plane put the metric where the shortest way from a point $P$ to a point $Q$ is to travel down a vertical line to the $x$-axis, go across the $x$-axis, and then up the vertical line to $Q$. On the closed lower half plane, put the usual Euclidean metric. Since the metrics agree on the $x$-axis (you just travel across the $x$-axis since there is no need to go up/down vertical lines), the gives a well-defined metric on the whole plane. $\endgroup$ – Robert Bell Apr 12 '18 at 16:17
  • 1
    $\begingroup$ The metric on the closed upper half plane is "tree-like," which is quite different from the Euclidean metric. Both this metric on the closed upper half plane and xyzzyz's example are examples of $\mathbb{R}-trees. xyzzyz's metric induces a topology which makes the closed upper half plane homeomorphic to a tree (1-dimensional CW complex); but the one I've suggested does not; it does not satisfy the "closure finiteness" condition of a CW complex. Anyway, this gives some context to where I first learned of this metric on the plane (or half plane). $\endgroup$ – Robert Bell Apr 12 '18 at 16:21

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.