7

Show that $f(x) = e^x$ is continuous using the epsilon-delta definition.

I can't seem to write down anything meaningful...

kiwifruit
  • 707
  • 2
    What is known about $e^x$, and how is it defined? – Daniel Fischer Mar 23 '14 at 00:33
  • I can see that is is continuous geometrically of course, but not sure how to get this formally – kiwifruit Mar 23 '14 at 00:34
  • 1
    How to prove it formally depends on what you already know, and which definition you have. The power series definition, $$e^x = \sum_{n=0}^\infty \frac{x^n}{n!},?$$ Do you have the addition theorem, $e^{x+y} = e^x\cdot e^y$? – Daniel Fischer Mar 23 '14 at 00:36
  • Hmm, this came up in an integral equation context when I was proving existence of the solution. I don't think I've seen the power series definition, but the addition theorem would apply – kiwifruit Mar 23 '14 at 00:37
  • 1
    Apply the definition of continuity everywhere: for all $c$ and for all $\epsilon > 0$, there exists $\delta > 0$ such that if $|x - c| < \delta$, then $|f(x) - f(c)| < \epsilon$. So since $e^{x}$ is defined everywhere on $\mathbb{R}$, let $c \in \mathbb{R}$ and let $\epsilon > 0$. Can you produce a corresponding $\delta > 0$? – ml0105 Mar 23 '14 at 00:38
  • I tried doing that, but got stuck simplifying $|e^x-e^c|$ – kiwifruit Mar 23 '14 at 00:40
  • 3
    $\lvert e^x-e^c\rvert = e^c\cdot\lvert e^{x-c}-1\rvert$. That reduces the problem to showing continuity in $0$. – Daniel Fischer Mar 23 '14 at 00:42
  • Hmm, and how can I apply the $|x-c|<\delta$ part here if I can't plug it in? – kiwifruit Mar 23 '14 at 00:47
  • where did the 1 come from in $\lvert e^x-e^c\rvert = e^c\cdot\lvert e^{x-c}-1\rvert$ – Zophikel Jul 26 '17 at 19:18

4 Answers4

7

Let $a$ be a positive real number. Then the function $f: \mathbb{R}\to \mathbb{R}$ defined by $x\mapsto a^x$ is continuous.

Proof:

1) First we prove the continuous at $0$.

We may assume that $a>1$. Let $\varepsilon>0$ be arbitrary. Since $a^{1/k}\to 1$ and $a^{-1/k}\to 1$ as $k\to \infty$, we choose $K$ such that both are $\varepsilon$-close to $1$. Let $\delta=1/K$, so for $|x|<1/K$ we have

\begin{align}-1/K<x<1/K\\a^{-1/K}<a^{x}<a^{1/K} \end{align}

which proves that $a^{x}$ is $\varepsilon$-close to $1$ as desired (the case $a\le1$ is handled similarly just with the inequality in the other direction).

2) To conclude we prove the continuity in the general case.

Let $x_0\in \mathbb{R}$, we have to show that $\lim_{x\to x_0}a^x=a^{x_0}$. Since $x-x_0\to 0$ as $x\to x_0$, then $a^{x-x_0}\to1$ by ($1$) and thus

$$\lim_{x\to x_0}a^x=\lim_{x\to x_0}a^{x_0}a^{x-x_0}=a^{x_0}\lim_{x\to x_0}a^{x-x_0}=a^{x_0}$$

as was to be shown.

Since $e\in \mathbb{R}^{>0}$, thus $e^x$ is continuous.

Jose Antonio
  • 7,154
  • For your proof in 1. to work, you need to show that $f$ is monotone, and this is non-trivial if you want to do it from scratch. – Matematleta Mar 09 '23 at 14:33
3

Start with $|x-c|<\delta$.

Let's take a look at the case that $x\ge c$. Then: $$x-c <\delta$$ $$x < c+ \delta$$ $$e^x < e^{c+\delta}$$ $$0 \le e^x-e^c < e^c(e^\delta - 1)$$ Let's pick $\delta$ such that $\varepsilon = e^c(e^\delta - 1)$. Then: $$\delta = \ln(1+\varepsilon e^{-c})$$ Repeat for $x<c$.

In other words, you are able to find a $\delta$ for any $\varepsilon>0$.

Qed.

  • The fact that the logarithm function is well defined requires one to show that $x \mapsto e^x$ is bijective. Injectivity is relatively easy (the guy is strictly increasing). However, surjectivity is usually shown by the fact that the map is continuous, which is what we are trying to prove here. I think that's circular logic. – Thomas Winckelman Dec 24 '19 at 03:54
2

Almost identical questions have been answered $n + 1$ times on this site. The proof is trivial. I will present a slight variation of the standard proof.

Recall that, by definition, $f(x)$ is continuous on $\mathbb{R}$ if and only if $$(\forall x_0 \in \mathbb{R})(\forall \varepsilon > 0)(\exists \delta > 0)(\forall x \in \mathbb{R})(0 < |x - x_0| < \delta \implies |f(x) - f(x_0)| < \varepsilon).$$

Combining a well-worn result $$e^x = \lim_{n \to \infty}\left(1 + \frac{x}{n}\right)^n,$$ and the fact that, by definition, $\lim_{x \to x_0} f(x) = L$ if and only if $$(\forall \varepsilon > 0)(\exists \delta > 0)(0 < |x - x_0| < \delta \implies |f(x) - L| < \varepsilon),$$ proves that $e^x$ is continuous.

Edit: To summarize, we know that the sequence $$s_n = \left(1 + \frac{x}{n}\right)^n$$ converges to $e^x$ as $n \to \infty$. We see that $$\lim_{x \to x_0} \lim_{n \to \infty} \left(1 + \frac{x}{n}\right)^n = \lim_{x \to x_0} e^x = e^{x_0} = \lim_{n \to \infty} \left(1 + \frac{x_0}{n}\right)^n = \lim_{n \to \infty} \lim_{x \to x_0} \left(1 + \frac{x}{n}\right)^n.$$ Moreover, by definition, $\lim_{x \to x_0} f(x) = L$ if and only if $$(\forall \varepsilon > 0)(\exists \delta > 0)(0 < |x - x_0| < \delta \implies |f(x) - L| < \varepsilon).$$ Yet, since $x_0$ is arbitrary, we have $$(\forall x_0 \in \mathbb{R})(\forall \varepsilon > 0)(\exists \delta > 0)(\forall x \in \mathbb{R})(0 < |x - x_0| < \delta \implies |f(x) - f(x_0)| < \varepsilon).$$ Thus, $e^x$ is continuous for all $x$.

glebovg
  • 10,154
  • Who downvoted my answer? Please elaborate. – glebovg Mar 23 '14 at 01:05
  • 7
    I didn't downvote you, but I think I'm missing something. Using your outline, one would need to show that $\lim_{x \to x_0} e^x = e^{x_0}$, or $\lim_{x \to x_0} \lim_{n \to \infty} (1 + \frac{x}{n})^n = \lim_{n \to \infty} (1 + \frac{x_0}{n})^n$, which amounts to justifying interchanging the limits, no? – anonymous Mar 23 '14 at 04:39
  • @anonymous Yes. I edited my answer to address your comment. – glebovg Mar 23 '14 at 07:06
  • 1
    Oh, ha, I see. Cute! – anonymous Mar 23 '14 at 15:40
1

As discussed in this forum, it suffices to show continuity at zero. Using the fact that $E(0)=1,$ we see that for any $\delta<1,$ every $x\in\mathbb{R}$ (or $x\in\mathbb{C}$) such that $|x|<\delta$ satisfies \begin{align*} \left|E(x) - E(0)\right| \hspace{1mm}=\hspace{1mm} \left|\sum_{n=1}^\infty \frac{x^n}{n!}\right| \hspace{1mm}\leq\hspace{1mm} \sum_{n=1}^\infty \frac{|x|^n}{n!} \hspace{1mm}<\hspace{1mm} \sum_{n=1}^\infty \frac{\delta^n}{n!} \hspace{1mm}<\hspace{1mm} \sum_{n=1}^\infty \delta^n \hspace{1mm}=\hspace{0.5mm} \frac{1}{1-\delta} - 1, \end{align*} at which point the reader is invited to choose $\delta$ in terms of a given $\varepsilon$.