0
$\begingroup$

In topology, continuity is defined as:

A function $f:X\rightarrow Y$ is continuous if the inverse image of an open set in $Y$ is an open set in $X$.

I have a problem to use it to check the non-continuous function. For example, in J.Munkres' book Topology (2nd Edition) (Pg.109), there is an example $$f(x)=\begin{cases} x-2, & x<0\\ x+2, & x\geq0 \end{cases}$$ The domain of this function (i.e. the $X$ in the definition) is $\mathbb{R}$, i.e. $x\in(-\infty,\infty)$; the codomain (i.e. the $Y$ in the definition) is $(-\infty,-2)\cup[2,\infty)$. To prove this function is discontinuous at $x=0$, the book choose the open set of $f$ as $(1,3)$, and thus the inverse image is $[0,1)$, which is not an open set.

The problem is:

the open set $(1,3)$ is not the subset of the codomain $(-\infty,-2)\cup[2,\infty)$. How can we choose it?

In my point of view, within the codomain $(-\infty,-2)\cup[2,\infty)$, all allowed open sets are either within the subset $(-\infty,-2)$ or within the subset $[2,\infty)$. Within these two parts, the function are always continuous. How can we prove the discontinuous?

$\endgroup$
13
  • 2
    $\begingroup$ You can, if no codomain is actually specified; it just makes things unnecessarily complicated. If you treat $Y=(\leftarrow,-2)\cup[2,\to)$ as the codomain, then you have to understand that it inherits its topology from $\Bbb R$, so $(1,3)\cap Y=[2,3)$ is an open subset of $Y$. $\endgroup$ Commented May 31, 2020 at 22:30
  • 2
    $\begingroup$ But $Y$ is an open set in the subspace $Y$ of $\Bbb R$. Recall that the subspace topology on $Y$ is $\{U\cap Y:U\text{ is open in }\Bbb R\}$. $\endgroup$ Commented May 31, 2020 at 22:40
  • 2
    $\begingroup$ If you insist on using $Y$ as the codomain instead of the natural one, $\Bbb R$, then you have to use $[2,3)$ as your open set in the codomain, not $(1,3)$. In practical terms it comes to exactly the same thing. $\endgroup$ Commented May 31, 2020 at 22:45
  • 2
    $\begingroup$ $[2,3)$ is an open set in $Y$ when $Y$ has the subspace topology that it inherits from $\Bbb R$. If you take $Y$ as the codomain, then you must work in that topology. If you want to use the topology of $\Bbb R$ itself, you must take $\Bbb R$ as the codomain. $\endgroup$ Commented May 31, 2020 at 22:54
  • 2
    $\begingroup$ Of course there are topologies on $Y$ in which $[2,3)$ is a closed set; the discrete topology is one of them. But none of them is at all relevant to the original problem. In that problem you are expected to understand that $f$ is a function from $\Bbb R$ to $\Bbb R$. If you artificially take its range $Y$ to be its codomain, then you are still required to use the relevant topology, which is the subspace topology that $Y$ inherits from $\Bbb R$. $\endgroup$ Commented May 31, 2020 at 22:58

1 Answer 1

2
$\begingroup$

The codomain of your function is $\Bbb R$, not $(-\infty,-2)\cup[2,\infty)$. So, there is no problem, since $(1,3)\subset\Bbb R$.

If you want to see $f$ as a map from $\Bbb R$ into $Y=(-\infty,-2)\cup[2,\infty)$, then, instead of $(1,3)$, take its intersection with$Y$, which is $[2,3)$. It is an open subset of $Y$, but $f^{-1}\bigl([2,3)\bigr)$ is not an open subset of $\Bbb R$.

$\endgroup$
4
  • $\begingroup$ I don't get it clearly. If we talk the intersection within $\mathbb{R}$, $Y=(-\infty, -2)\cup[2,\infty)$ itself is not an open set in $\mathbb{R}$, how can the intersection to be an open set? If we talk the intersection within $Y$ (then $Y$ is an open set), but the $(1,3)$ is out of bounds (is not within $Y$), how can we make this intersection? $\endgroup$ Commented May 31, 2020 at 22:44
  • 1
    $\begingroup$ By definition, a subset $A$ of $Y$ is an open set if there is an open subset $A^\star$ of $\Bbb R$ such that $A=Y\cap A^\star$. And $[2,3)=Y\cap(1,3)$. $\endgroup$ Commented May 31, 2020 at 22:48
  • $\begingroup$ Can you say it a little more clear? This is the definition of inherited topology? I think we can definitely treat $Y$ isolated to a bigger set. And define a topology with $[2,3)$ a close set? $\endgroup$ Commented May 31, 2020 at 22:58
  • 2
    $\begingroup$ Yes, what I wrote is the definition of inherited topology (although the usual name is subspace topology). $\endgroup$ Commented May 31, 2020 at 23:00

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.