Existence of a continuous function which does not achieve a maximum.

Question

Suppose $X$ is a non-compact metric space. Show that there exists a continuous function $f: X \rightarrow \mathbb{R}$ such that $f$ does not achieve a maximum.

I proved this assertion as follows:

If $X$ is non-compact, there exists a sequence which has no converging subsequence. Therefore, the set $E$ of elements of this sequence is infinite and has no limit points. Therefore, in the induced topology, $E$ is a countable discrete space. Let $e_n$ be a enumeration of $E$. Define $f: E \rightarrow \mathbb{R}$ as $f(e_n)=n$. Since $E$ is discrete, $f$ is continuous. By Tietze (note that $E$ is closed in $X$), there exists a continuous function $g: X \rightarrow \mathbb{R}$ which extends $f$. But $f$ is unbounded, then so is $g$, and $g$ does not achieve a maximum. $\square$

My first question is: Is everything in the proof fine?

Now, my second question is: Can you provide me a proof which is more elementary? (Not using Tietze, for example. This exercise is just after the definition of compactness. I suppose this proof is not what was intended.)

And my third question is: Does the proposition hold for arbitrary topological spaces? If not, can you provide a counter-example?

Answer 1

Here is an explicit and elementary construction, using only the fact that a metric space is compact iff every sequence has an accumulation point. Let $(x_n)$ be a sequence in $X$ with no accumulation point, and define $g(x)=\inf_n d(x,x_n)+1/n$ and $f(x)=1/g(x)$. It is easy to see that $g(x)=0$ iff $x$ is an accumulation point of $(x_n)$, so $g$ is never $0$ and $f$ is defined on all of $X$. Moreover, $f(x_n)\geq n$, so $f$ is unbounded. Finally, it is straightforward to check that $g$ is continuous, and hence so is $f$.

Answer 2

Your argument is correct and is probably the most straightforward. The result does not even hold in all $T_3$ spaces: in this answer I give a general method, due to Eric van Douwen, for starting with a $T_3$ space having two points that cannot be separated by a continuous real-valued function and producing from it a $T_3$ space on which all continuous real-valued functions are constant. The other answer to that question gives references to two $T_3$ spaces having two points that cannot be separated by a continuous real-valued function, and this answer expands on one of those spaces and gives yet another example of such a space.

Answer 3

The proof seems fine but Tietze needs to have the function continuous on a closed subset, but as it has no limit points this is given. If your function Needs to be continuous I can tell you the third question to have a negative answer.

In fact there is a non compact Hausdorff space such that the continuous functions are exactly the constant ones. Take $\mathbb{Z}^+$ with the relatively prime integer topology, here it is the case that for every non empty open sets $U,V$ the intersection $\overline{U}\cap \overline{V}\neq \varnothing$ hence every continuous function to $\mathbb{R}$ must be constant and so surely attains a Maximum. To see it is not compact is a bit more difficult, here you use that if it is compact it needs to be $T_3$ and $T_4$ (as it is Hausdorff), but with $T_4$ we would have Urysohn-Lemma which grants you more continuous mappings.