Let $~f: \mathbb{R} \to \mathbb{R}$ be bounded and uniformly continuous. Show that $e^f$ is also bounded and uniformly continuous.

My approach (with 5xum's extensive help) goes as follows:

  1. $f$ bounded $\implies$ $e^f$ bounded

Since there exists an $M \in \mathbb{R}_{>0}$ so that $|f(x)| \leq M$ for all $x \in \mathbb{R}$, by the continuity and monotonicity of $e^x: \mathbb{R} \to \mathbb{R}$, we can see that $|e^{f(x)}| = e^{f(x)} \leq e^M ~\forall x \in \mathbb{R}$.

  1. $f$ uniformly continouos. $\implies$ $e^f$ uniformly continuous.

We want to show that $$ \forall \varepsilon' > 0 ~\exists \delta' > 0 ~\forall x,y \in \mathbb{R}: (|x-y| < \delta' \implies |e^{f(x)} - e^{f(y)}| < \varepsilon') $$ Let $\varepsilon' > 0$ be fixed and set $\varepsilon := \frac{\varepsilon'}{2 M} > 0$, where $M$ is the bound of $\max\{e^{f(x)},e^{f(y)}\}$, meaning $|e^{f(x)}|, |e^{f(y)}| \leq M ~\forall x,y \in \mathbb{R}~~(*_2)$.

Since $f$ is uniformly continuous, the following holds $$ \forall \varepsilon > 0 ~\exists \delta > 0 ~\forall x,y \in \mathbb{R}: |x - y| < \delta \implies |f(x) - f(y)| < \varepsilon ~~(*_1) $$

Now, let $\delta > 0$ be so that $|x - y| < \delta$ and without loss of generality let $f(y) \geq f(x) ~\forall x,y\in\mathbb{R}$. Then we obtain $$ \begin{align*} |e^{f(x)} - e^{f(y)}| = |e^{f(x)}| |1 - e^{f(y) - f(x)}| &= e^{f(x)} |e^{|f(y) - f(x)|} - 1| \\ &\overset{(*)}{<} 2 e^{f(x)} |f(x) - f(y)| \\ &\overset{(*_1)}{<} 2 e^{f(x)} \varepsilon \\ &\overset{(*_2)}{\leq} 2 \varepsilon M \\ &=\varepsilon' \end{align*} $$

Where $(*)$ can be done because $|e^u - 1| < 2u ~~\forall u \in (0,1]$.


1 Answer 1

  1. $f$ bounded $\implies$ $e^f$ bounded

trivial because of the monotonicity of $e^x$ ??

My professor had a favourite saying: "Dear colleague, 'trivial' and 'obvious' are my words. They are not yours to use."

If what you wrote is so trivial, then you should be able to prove it in $3$ lines, and if you are able to prove in $3$ lines, then just do it.

Beside, $f(x)=\frac1{x-1}$ is a monotonic function on $(0,1)$, but it is not bounded, so your argument obviously needs work.

  1. $f$ uniformly cts. $\implies$ $e^f$ uniformly cts.

We want to show that $$ \forall \varepsilon' > 0 ~\exists \delta' > 0 > \forall x,y \in \mathbb{R}: (|x-y| < \delta' \implies |f(x) - f(y)| < > \varepsilon') $$

No. We assume that what you wrote is true. What we need to prove is the following:

$$\forall\epsilon >0 \exists \delta > 0 \forall x,y\in\mathbb R:(|x-y|<\delta\implies |e^{f(x)} - e^{f(y)}| < \epsilon)$$

You need to prove the uniform continuity of $e^f$, while assuming that $f$ is uniformly continuous and bounded.

After your edit:

Part 1 is ok now.

Part 2 is still strange.

First problem:

Let $\varepsilon > 0$ be fixed and $\varepsilon' > 2 M \varepsilon >0$

This is very strange. Your proof should start with "Let $\varepsilon' > 0$", not "$\varepsilon > 0$." I think this is only a consequence of sloppy writing, though, so you should write, instead of the above,

Let $\varepsilon' > 0$ be fixed and set $\varepsilon = \frac{\varepsilon'}{2M}$

And then the proof should work out.

Second problem:

From $|x - y| < \delta$ follows that $|f(x) - f(y)| < \varepsilon$ because $f$ is uniformly cts.

This is very sloppy writing. What is $\delta$ equal to here? You haven't defined $\delta$, it just appeared out of the blue. This is another example of sloppy writing that must be fixed.

Third problem:

$$|e^{f(x)} - e^{f(y)}| = |e^{f(x)}| |e^{|f(x) - f(y)|} - 1|$$

This is false if $f(x)<f(y)$. For example, if $f(x)=0$ and $f(y)=1$, then the left side of the equation equals $|e^0-e^1|=e$, while the right side equals $|e^0||e^{|0-1|} - 1| = 1\cdot |e-1|=e-1$

Fourth problem:

$$|e^{f(x)}| |e^{|f(x) - f(y)|} - 1| < e^{f(x)} |e^{\varepsilon} - 1|$$

How do you know this inequality is true? If I were you, I would first use the knowledge you write in $*$ before going for the estimate with $\delta$...

After your second edit:

Slight problem:

Don't say "let $\delta$ be fixed", better to just say "let $\delta$ be such that if $|x-y|<\delta$, then $|f(x)-f(y)|<\epsilon$. That way, it's clear you only fix one number ($\epsilon$) and then "calculate" what $\delta$ is.

Bigger mistake:

You still wrote

$$|e^{f(x)}| |1 - e^{f(y) - f(x)}| = e^{f(x)} |e^{|f(y) - f(x)|} - 1|$$

which is not true, since, for $f(x)=1$ and $f(y)=0$, the left side equals $$|e^1|\cdot |1-e^{0-1}| = e\cdot (1-e^-1) = e-1$$

while the right side equals

$$|e^1|\cdot |e^{|0-1| -1} = e\cdot (e-1)=e^2-1$$

  • $\begingroup$ I wasn't sure how to prove the boundness (hence the ? at the end), the 'trivial'-statement was a hint received from a friend. I now added a proof to my question, is it correct? And you're right about the "We want to show"-part. I misspelled and now edited my question. $\endgroup$
    – Ramanujan
    Apr 10, 2018 at 12:03
  • 1
    $\begingroup$ @ViktorGlombik Your proof is very very sloppy. You obviously have the correct idea, but if I was grading this, you would struggle to get half the points. There are at least 4 problems with your proof which I described in my edited answer. $\endgroup$
    – 5xum
    Apr 10, 2018 at 12:22
  • $\begingroup$ I've tried to incorporate your suggestions. Is my proof satisfactory now? $\endgroup$
    – Ramanujan
    Apr 10, 2018 at 13:00
  • $\begingroup$ @ViktorGlombik Unfortunatelly, no. It's much closer, but still sloppy. $\endgroup$
    – 5xum
    Apr 10, 2018 at 13:09
  • $\begingroup$ but I said that log $f(y) \geq f(x)$, so that $f(y) - f(x) = |f(y) - f(x)|$. In your counterexample, you ignore that condition, or am I mistaken? $\endgroup$
    – Ramanujan
    Apr 10, 2018 at 13:29

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.