Simplest Or Nicest Proof That $1+x \le E^x - Math Stack Exchange

1. 1. Home
   2. Questions
   3. Tags
   4. Users
   5. Unanswered
2. Teams

   Ask questions, find answers and collaborate at work with Stack Overflow for Teams.

   Try Teams for free Explore Teams
3. Teams

4. Ask questions, find answers and collaborate at work with Stack Overflow for Teams. Explore Teams

Teams

Q&A for work

Connect and share knowledge within a single location that is structured and easy to search.

Learn more about Teams Simplest or nicest proof that $1+x \le e^x$ Ask Question Asked 11 years, 3 months ago Modified 2 years, 7 months ago Viewed 94k times 128 $\begingroup$

The elementary but very useful inequality that $1+x \le e^x$ for all real $x$ has a number of different proofs, some of which can be found online. But is there a particularly slick, intuitive or canonical proof? I would ideally like a proof which fits into a few lines, is accessible to students with limited calculus experience, and does not involve too much analysis of different cases.

• inequality
• exponential-function
• big-list

Share Cite Follow edited Feb 17, 2018 at 10:41 Parcly Taxel 105k2020 gold badges116116 silver badges202202 bronze badges asked Sep 25, 2013 at 10:04 Ashley MontanaroAshley Montanaro 1,39722 gold badges99 silver badges88 bronze badges $\endgroup$ 3

• 2 $\begingroup$ Question about the same inequality: Prove that $e^x\ge x+1$ for all real $x$ $\endgroup$ – Martin Sleziak Commented Sep 25, 2013 at 12:31
• 2 $\begingroup$ There are many great answers below. I now wonder whether, as a question with no objective right answer, this should be community wiki? $\endgroup$ – Ashley Montanaro Commented Sep 26, 2013 at 21:47
• $\begingroup$ This is a duplicate of this : math.stackexchange.com/questions/252541/… $\endgroup$ – Guy Fsone Commented Nov 10, 2017 at 18:30

Add a comment |

26 Answers 26

Sorted by: Reset to default Highest score (default) Date modified (newest first) Date created (oldest first) 181 $\begingroup$

Another way (not sure if its "simple" though!): $y = x+1$ is the tangent line to $y = e^x$ when $x= 0$. Since $e^x$ is convex, it always remains above its tangent lines.

Share Cite Follow answered Sep 25, 2013 at 10:23 MacavityMacavity 47.5k66 gold badges3838 silver badges7272 bronze badges $\endgroup$ 8

• 15 $\begingroup$ That's my favourite argument. I'm afraid it fails on the "is accessible to students with limited calculus experience" criterion, but it's boootiful. $\endgroup$ – Daniel Fischer Commented Sep 25, 2013 at 12:28
• 19 $\begingroup$ @DanielFischer: Are you sure that it fails? Tangent lines can appear much earlier then the derivatives and it have a nice graphical representation (sure, representation might not be mathematically precise). $\endgroup$ – Maja Piechotka Commented Sep 25, 2013 at 13:38
• 4 $\begingroup$ @Maciej I'm not sure. But without some calculus to build on, I think it would be too hand-wavy for me. $\endgroup$ – Daniel Fischer Commented Sep 25, 2013 at 13:44
• 20 $\begingroup$ The convexity, at least, is elementary: $e^{x+z} - 2 e^x + e^{x-z} = (e^z-1)^2 e^{x-z} > 0$ for all real $x,z$, with equality iff $z=0$. $\endgroup$ – Noam D. Elkies Commented Sep 25, 2013 at 14:55
• 30 $\begingroup$ @NoamD.Elkies -- for large values of "elementary". $\endgroup$ – Michael Lorton Commented Sep 25, 2013 at 18:52

| Show 3 more comments 101 $\begingroup$

$$ e^x = \lim_{n\to\infty}\left(1+\frac xn\right)^n\ge1+x $$

by Bernoulli's inequality.

Share Cite Follow answered Sep 25, 2013 at 14:21 John GowersJohn Gowers 25.4k66 gold badges6666 silver badges102102 bronze badges $\endgroup$ 1

• 2 $\begingroup$ This is what I thought of when I saw this question. (+1) $\endgroup$ – robjohn ♦ Commented Sep 27, 2013 at 0:31

Add a comment | 74 $\begingroup$

The shortest proof I could think of: $$1 + x \leq 1 + x + \frac{x^2}{2!} + \frac{x^3}{3!} + \cdots = e^x.$$

However, it is not completely obvious for negative $x$.

Using derivatives:

Take $f(x) = e^x - 1 - x$. Then $f'(x) = e^x - 1$ with $f'(x) = 0$ if and only if $x = 0$. But this is a minimum (global in this case) since $f''(0) = 1 > 0$ (the second derivative test). So $f(x) \geq 0$ for all real $x$, and the result follows.

Another fairly simple proof (but it uses Newton's generalization of the Binomial Theorem which is often covered in precalculus):

We proceed by contradiction. Suppose the inequality does not hold, i.e., $e^x < 1 + x$ for some $x$. Then $e^{kx} < (1 + x)^k$. Now set $x = 1/k$ so that \begin{align*} e &< \left( 1 + \frac{1}{k} \right)^k\\ &= 1 + \frac{k}{1}\left( \frac{1}{k} \right)^1 + \frac{k(k - 1)}{1 \cdot 2}\left( \frac{1}{k} \right)^2 + \frac{k(k - 1)(k - 2)}{1 \cdot 2 \cdot 3}\left( \frac{1}{k} \right)^3 + \cdots\\ &< 1 + \frac{k}{1}\left( \frac{1}{k} \right)^1 + \frac{k^2}{1 \cdot 2}\left( \frac{1}{k} \right)^2 + \frac{k^3}{1 \cdot 2 \cdot 3}\left( \frac{1}{k} \right)^3 + \cdots\\ &= 1 + 1 + \frac{1}{2!} + \frac{1}{3!} + \cdots\\ &= e, \end{align*} which is absurd. Therefore $1 + x \leq e^x$ for all real $x$.

By the way, this is where $$e = \lim_{k \to \infty}\left( 1 + \frac{1}{k} \right)^k$$ comes from because $$\lim_{k \to \infty}\frac{k(k - 1)}{k^2} = \lim_{k \to \infty}\frac{k(k - 1)(k - 2)}{k^3} = \cdots = \lim_{k \to \infty}\frac{k(k - 1)(k - 2) \cdots (k - n)}{k^{n + 1}} = \cdots = 1.$$

Share Cite Follow edited Sep 26, 2013 at 17:49 answered Sep 25, 2013 at 10:13 glebovgglebovg 10.3k22 gold badges3535 silver badges5858 bronze badges $\endgroup$ 5

• 23 $\begingroup$ Is it that evident for negative $x$? $\endgroup$ – Macavity Commented Sep 25, 2013 at 10:16
• 4 $\begingroup$ @Macavity for x<-1 is obvious $\endgroup$ – BЈовић Commented Sep 25, 2013 at 17:35
• 1 $\begingroup$ In your third proof : Why should such a $x$ be of the form $1/k$ with $k$ a positive integer ? $\endgroup$ – user37238 Commented Sep 26, 2013 at 15:20
• 1 $\begingroup$ @user37238 Here, $k$ is not necessarily an integer because, technically, the proof uses the Binomial Series. This explains why I wrote the binomial coefficients using the falling factorial. $\endgroup$ – glebovg Commented Sep 26, 2013 at 17:58
• $\begingroup$ I'm confused: you assume (for a contradiction) that $e^{x}<1+x$ for some real $x$, but then you claim that you can set $x=1/k$ without further justification. Would you not need to prove that $1/k$ is a valid choice of $x$ under such an assumption? $\endgroup$ – Will R Commented Jul 28, 2015 at 8:52

Add a comment | 51 $\begingroup$

Let $f(x) = e^x-(1+x)$, then $f^\prime(x) = e^x-1$. Hence $f^\prime(x)=0$ iff $x=0$. Furthermore $f^{\prime\prime}(0) = e^0=1>0$, thus $f(0)=0$ must be the global minimum of $f$, proving your claim.

Share Cite Follow answered Sep 25, 2013 at 10:14 AbelAbel 7,3921919 silver badges2222 bronze badges $\endgroup$ 1

• 2 $\begingroup$ I like the idea behind this proof : just study a function and show that this function is always non-negative. But I think that it could be a little simpler if you say that $f'\ge 0$ on $[0,+\infty($ (and consequently $f$ increases on this interval) and $f'\le 0$ on $)-\infty,0]$ (and $f$ decreases on this interval) so $f(0)=0$ is a global minimum of $f$. $\endgroup$ – user37238 Commented Sep 25, 2013 at 14:59

Add a comment | 39 $\begingroup$

One that's not been mentioned so far(?): knowing that $$ 0 < e^x = 1 + x + \frac{x^2}{2} + \frac{x^3}{3!} + \frac{x^4}{4!} + \frac{x^5}{5!} + \cdots $$ proves the inequality except for $-1 < x < 0$. But in that region $$ e^x - (1+x) = \frac{x^2}{2} + \frac{x^3}{3!} + \frac{x^4}{4!} + \frac{x^5}{5!} + \cdots $$ is an alternating series whose terms decrease in absolute value and start out positive. Therefore it is positive by the usual argument: group the terms as $$ e^x - (1+x) = \left( \frac{x^2}{2} + \frac{x^3}{3!} \right) + \left( \frac{x^4}{4!} + \frac{x^5}{5!} \right) + \cdots $$ and observe that each combined term is positive, QED.

(This actually works for $-3 < x < 0$, but you still want to use $e^x > 0$ to prove the inequality for very negative $x$.)

Share Cite Follow edited Sep 26, 2013 at 20:51 answered Sep 25, 2013 at 14:54 Noam D. ElkiesNoam D. Elkies 26.2k11 gold badge6767 silver badges8383 bronze badges $\endgroup$ 2

• 2 $\begingroup$ Why does knowing that $0<e^x$ give the result for $x\le-1$? $\endgroup$ – John Gowers Commented Sep 25, 2013 at 14:56
• 15 $\begingroup$ $x \leq -1$ means $x+1 \leq 0 < e^x$. $\endgroup$ – Noam D. Elkies Commented Sep 25, 2013 at 14:58

Add a comment | 27 $\begingroup$

Repeatedly using $1 + x \le \left(1 + \frac{x}{2} \right)^2$, we have \begin{align} 1 + x \le \left(1 + \frac x 2\right)^2 \le \left(1 + \frac x 4\right)^4 \le \left(1 + \frac x 8\right)^8 \le \dots \le \left(1 + \frac x {2^k}\right)^{2^k}. \end{align} Taking the limit of $k \rightarrow \infty$ yields $$ 1 + x \le e^x. \qquad\qquad(1) $$

Another proof using the technique in this post. By the AM-GM inequality, $$ \sqrt[n]{1 \times \cdots \times 1 \times (1 + x)} \le \frac{1 + \dots + 1 + (1 + x)}{n} =1 + \frac{x}{n}. $$ So, $$ 1+x \le \left(1 + \frac{x}{n} \right)^n. $$ Taking the limit of $n \rightarrow \infty$ yields (1).

Share Cite Follow edited Apr 13, 2017 at 12:21 community wiki 3 revshbp $\endgroup$ Add a comment | 21 $\begingroup$

If $x \ge 0$ then $$\begin{align} e^x = 1 + \int_0^x e^t\,\mathrm dt &= 1 + \int_0^x\left( 1 + \int_0^t e^u \,\mathrm du\right)\,\mathrm dt \\&= 1+x + \int_0^x \int_0^t e^u\,\mathrm du\,\mathrm dt \ge 1+x\end{align}$$

If $x \le 0$ then $$\begin{align}e^x = 1 - \int_x^0 e^t\,\mathrm dt &= 1 - \int_x^0\left( 1 - \int_t^0 e^u\,\mathrm du\right)\,\mathrm dt \\ &= 1+x + \int_x^0 \int_t^0 e^u\,\mathrm du\,\mathrm dt \ge 1+x\end{align}$$

Share Cite Follow edited Sep 26, 2013 at 12:12 filmor 55122 silver badges99 bronze badges answered Sep 25, 2013 at 10:45 merciomercio 50.9k22 gold badges8383 silver badges133133 bronze badges $\endgroup$ Add a comment | 14 $\begingroup$

For completeness, using $\exp(x)=1+x+\frac{1}{2}x^2+\dots$, the inequality is trivial for $x\ge 0$. It is also trivial for $x<-1$.

It remains to show the case $-1<x<0$. Replacing $x$ by $-x$, one need to show $1-x < e^{-x}$ for $0<x<1$, or $$1+x+\frac{1}{2}x^2+\dots=e^x <\frac{1}{1-x}=1+x+x^2+\dots,$$ we are done.

Share Cite Follow edited Jun 23, 2019 at 13:12 J. W. Tanner 63k44 gold badges4242 silver badges8888 bronze badges answered Sep 25, 2013 at 11:41 Ma MingMa Ming 7,5022020 silver badges3535 bronze badges $\endgroup$ 3

• $\begingroup$ @Dr.MV $x+1<0< e^x$ $\endgroup$ – Ma Ming Commented Nov 21, 2016 at 2:47
• $\begingroup$ Yes, of course. From the post, it seemed that you were inferring this from the series expansion. $\endgroup$ – Mark Viola Commented Nov 21, 2016 at 3:24
• $\begingroup$ Did you mean it's trivial for $x \mathbf \le -1$ ? $\endgroup$ – J. W. Tanner Commented Mar 14, 2019 at 0:31

Add a comment | 13 $\begingroup$

Beautiful answers, but nobody used The Mean Value Theorem. Apply MVT on $[0,x] $ for $x>0$. There is some $c\in (0,x)$ such that:

\begin{align} \frac{e^x-e^0}{x-0} = e^c > 1 \end{align} So \begin{align} e^x>1+x \end{align} Something similar can be done for $x<0$. Finally note that we have equality when $x=0$. So we get the desired result: \begin{align} e^x\geq 1+x \end{align}

Share Cite Follow answered Nov 10, 2017 at 19:57 ShashiShashi 8,89811 gold badge1414 silver badges3939 bronze badges $\endgroup$ Add a comment | 10 $\begingroup$

Completing glebovg's answer :

• the inequality $1+x \le e^x$ clearly holds for $x \leq -1$,

• suppose $x \geq -1$ :

the series $e^x = 1 + x + \frac{x^2}{2!} + \frac{x^3}{3!} + \dots$ can be written (grouping the terms in pair) :

$$e^x = 1 + x + \sum_{k \geq 1} \left( \frac{x^{2k}}{(2k)!} + \frac{x^{2k+1}}{(2k+1)!} \right)$$

$$e^x = 1 + x + \sum_{k \geq 1} x^{2k}\left( \frac{1}{(2k)!} + \frac{x}{(2k+1)!} \right)$$

$$e^x = 1 + x + \sum_{k \geq 1} x^{2k}\left( \frac{2k + 1 + x}{(2k+1)!} \right)$$

under the assumption $x \geq -1$, the $\sum$ part is clearly a sum of positive numbers.

Share Cite Follow edited Apr 13, 2017 at 12:19 CommunityBot 1 answered Sep 25, 2013 at 14:58 LeGECLeGEC 20111 silver badge33 bronze badges $\endgroup$ Add a comment | 7 $\begingroup$

We want to prove that $1+x\le e^x$ for any $x\in\mathbb R$. Setting $x=\log(u)$, this is equivalent to proving:

$$ 1+\log(u)\le u $$ for any $u\in (0, \infty)$.

This is true because:

$$ 1+\log(u)=1+\int_1^u\frac1tdt\le1+\int_1^u1dt=1+u-1=u $$

Some care is needed to establish that the inequality is true for both $u\ge1$ and $0<u\le1$. In the second case, we can see this more clearly by writing:

$$ 1+\int_1^u\frac1tdt=1+\int_u^1-\frac1tdt\le1+\int_u^1-1dt=1-1+u=u $$

Share Cite Follow edited Sep 26, 2013 at 11:55 answered Sep 25, 2013 at 15:20 John GowersJohn Gowers 25.4k66 gold badges6666 silver badges102102 bronze badges $\endgroup$ Add a comment | 6 $\begingroup$

There is an amusing proof that I found yesterday that $e^x>x$ for every $x\in \mathbb{R}$. It is obvious that $e^x>x$ if $x<0$ since the LHS is positive and the RHS is negative. Suppose that for some $a\ge 0$, the inequality $e^a\le a$ holds. Then $a\ge e^a\ge 1$ since $e^a\ge e^0$ because $a\ge 0$. But now we can see that $a\ge 1$ and again, $a\ge e^a\ge e^1$ and so $a\ge e$. We continue applying the same observation and conclude that $a\ge e^{^{e}}$ and so on, which means that $a$ is unbounded which is a contradiction.

Share Cite Follow edited Dec 6, 2020 at 17:35 answered May 21, 2019 at 9:57 Konstantinos GaitanasKonstantinos Gaitanas 9,19344 gold badges3131 silver badges4747 bronze badges $\endgroup$ 1

• 1 $\begingroup$ It's actually a great proof $\endgroup$ – emandret Commented Jan 17, 2021 at 0:24

Add a comment | 5 $\begingroup$

For positive values ​​of $ x $ We can use the following characterization of $e^x$ $$ e^x=\lim_{t\to \infty} \Big( 1+\frac{1}{t}\Big)^{tx},\quad t> 0,x\geq 0. $$ The Bernoulli's inequality states that $(1 + y)^r \geq 1 + ry$ for every $r \geq 1$ and every real number $y \geq −1$. Then for $y=\frac{1}{t}$ and $t>0$ such that $r=tx\geq 1$ we have \begin{align} e^x= &\lim_{t\to \infty} \Big( 1+\frac{1}{t}\Big)^{tx}\\ \geq &\lim_{t\to \infty} \Big(1+\frac{1}{t}(tx) \Big)\\ = & 1+x \end{align}

Share Cite Follow edited Dec 6, 2021 at 12:09 answered Sep 25, 2013 at 14:08 Elias CostaElias Costa 15.1k55 gold badges5555 silver badges9191 bronze badges $\endgroup$ 2

• $\begingroup$ I just came across this answer 8 years later haha. However, I believe that $(1+y)^r\geq 1+ry$ does not hold for $0 < r < 1$. For example, $(1+3)^{0.5} = 2$ while $1+3\cdot 0.5 = 2.5$. $\endgroup$ – Gareth Ma Commented Dec 6, 2021 at 8:30
• $\begingroup$ @GarethMa Yes, you are correct. The inequality is valid for $r \geq 1$. I will make the correction. $\endgroup$ – Elias Costa Commented Dec 6, 2021 at 12:08

Add a comment | 4 $\begingroup$

For $x>0$ we have $e^t>1$ for $0<t<x$

Hence, $$x=\int_0^x1dt \color{red}{\le} \int_0^xe^tdt =e^x-1 \implies 1+x\le e^x$$ For $x<0$ we have $e^{t} <1$ for $x <t<0$

$$-x=\int^0_x1dt \color{red}{\ge} \int^0_xe^tdt =1-e^x \implies 1+x\le e^x$$

Share Cite Follow edited Nov 10, 2017 at 18:38 answered Nov 10, 2017 at 18:28 Guy FsoneGuy Fsone 24.6k55 gold badges6565 silver badges108108 bronze badges $\endgroup$ Add a comment | 4 $\begingroup$

One which uses $\exp(x) = \frac 1{\exp(-x)} $ $$ 1 + x \underset{ \text{obvious}\\ \text{for $x>0$}}{\lt} 1 + x + {x^2 \over 2!} + {x^3 \over 3!} + ... = {1 \over 1 - x + {x^2 \over 2!} - { x^31 \over 3!} + ... } \tag 1 $$ Now we replace $+x$ by its negative counterparts and get similarily $$ 1 - x \underset{ \quad \text{for $x>0$}\\ \text{but not obvious}}{\lt} 1 - x + {x^2 \over 2!} - {x^3 \over 3!} + ... = {1 \over 1 + x + {x^2 \over 2!} + {x^3 \over 3!} + ... }\tag 2$$ But now the comparision with the fraction on the rhs becomes obvious if we look at the reciprocals. The reciprocal ${1\over 1-x}=1+x+x^2+x^3+...$ is and we get $$ {1 \over 1 - x} = 1+x+x^2+... \underset{ \text{obvious}\\ \text{for $x>0$}}{\gt} 1 + x + {x^2 \over 2!} + {x^3 \over 3!} + ... = {1 \over 1 - x + {x^2 \over 2!} - {x^3 \over 3!} + ... }\\ \tag 3 $$

Share Cite Follow edited Dec 16, 2021 at 16:47 answered Sep 26, 2013 at 8:52 Gottfried HelmsGottfried Helms 35.3k33 gold badges6868 silver badges145145 bronze badges $\endgroup$ 1

• 1 $\begingroup$ upps, I see there was another answer earlier, but which seemed too short for me to not only skim over it... $\endgroup$ – Gottfried Helms Commented Sep 26, 2013 at 9:28

Add a comment | 3 $\begingroup$

Proof by induction (works for natural numbers)

Assume it works for n

1 + n < e^n

Then we prove that it works for n+1

1 + (n+1) < e^(n+1)

Proof

1 + n < e^n or 1 + n + 1 < e^n + 1 or 1 + n + 1 < e^n + e^n since e^n > 1 or 1 + n + 1 < e^n * 2 or 1 + (n+1) < e^n * e since e > 2 or 1 + (n+1) < e^(n+1)

hence it is true for n+1 if true for n. We know it is true for 1, hence by induction is true of 2, 3, 4...so on.

Share Cite Follow edited Sep 25, 2013 at 23:19 answered Sep 25, 2013 at 22:41 tihomtihom 14733 bronze badges $\endgroup$ 4

• $\begingroup$ This isn't really a proof. Like, where is your inductive step? $\endgroup$ – Alexander Commented Sep 25, 2013 at 23:00
• 1 $\begingroup$ updated the answer works for natural number only though $\endgroup$ – tihom Commented Sep 25, 2013 at 23:20
• $\begingroup$ @tihom You can get a general proof from your proof in the following way: $floor(x)+1>x\geq floor(x) $ where $floor(x)$ is the greatest integer function. Now $$e^x\geq e^{floor(x)}> floor(x)+1> x$$ whenever $x\geq 1$ $\endgroup$ – Sedergine Commented Sep 18, 2021 at 2:07
• $\begingroup$ More precisely, this was for $e^x>x$ but refining your inductive argument to $e^n>n+2$ for $n\geq 2$ we can easily fix it. $\endgroup$ – Sedergine Commented Sep 18, 2021 at 2:34

Add a comment | 3 $\begingroup$

Another simple proof...

Define function $f(x)=e^x-(x+1)$. The minimum value is $0$ at $x=0$, it's also convex, so $f(x) \ge 0$.

Share Cite Follow answered Feb 1, 2018 at 15:43 karakfakarakfa 2,7151515 silver badges1515 bronze badges $\endgroup$ Add a comment | 2 $\begingroup$

For $x > 0$, consider the mean value theorem on the interval $[0,x]$. Then $$e^x - e^0 \geq \inf_{(0,x)} e^c \cdot (x - 0) = x,$$ implying $e^x \geq 1+x$. For $x < 0$, apply MTV on $[x,0]$: $$e^0 - e^x \leq \sup_{(x,0)} e^c \cdot (0 - x) = -x,$$ giving us $-e^x \leq -x - 1$, or $e^x \geq x + 1$. Then check $x = 0$ and the equality is proven.

Share Cite Follow answered Jul 28, 2015 at 7:28 user217285user217285 5,81511 gold badge1818 silver badges3131 bronze badges $\endgroup$ Add a comment | 1 $\begingroup$

The series expansion of $(1+x)$ is $(1+x)$, while $\exp(x)=1+x+\frac{1}{2}x^2+\frac{1}{6}x^3+...$. Subtracting the second from the first one you have the difference $d=\exp(x)-(1+x)=\frac{1}{2}x^2+\frac{1}{6}x^3+...$ which is zero only for $x=0$ otherwise $d\gt 0$ Q.E.D.

Share Cite Follow answered Sep 25, 2013 at 10:12 Riccardo.AlestraRiccardo.Alestra 10.7k44 gold badges4141 silver badges8484 bronze badges $\endgroup$ 3

• 3 $\begingroup$ Same comment, is it trivially evident for negative $x$ that $d>0$? $\endgroup$ – Macavity Commented Sep 25, 2013 at 10:17
• $\begingroup$ @Macavity: make the substitution: $x\to -y$. Anyway it's not trivial. $\endgroup$ – Riccardo.Alestra Commented Sep 25, 2013 at 10:26
• $\begingroup$ I have seen that proof done considering three regions separately - viz. $x \ge 0, -1 < x < 0, x \le -1$. The first region is trivial, the rest two not so much :( $\endgroup$ – Macavity Commented Sep 25, 2013 at 10:34

Add a comment | 1 $\begingroup$

Let $f(x)=\exp(x)-x-1$. Then, $f'(0)=0$. But $f$ is strictly convex (a difference of a strictly convex function and an affine one), so that $0$ most be a unique global minimum. Hence, $\exp(x)-x-1=f(x)\geq f(0)=0$ for all $x\in\mathbb{R}$.

Share Cite Follow answered Sep 25, 2013 at 10:35 triple_sectriple_sec 23.5k33 gold badges4343 silver badges9494 bronze badges $\endgroup$ Add a comment | 1 $\begingroup$

We want to show that (1) $$1+x\leq e^x,$$ for $x\in\mathbb{R}$. When $x\geq 0$, we have $$1+x\leq 1+x+\frac{x^2}{2!}+\frac{x^3}{3!}+\cdots=e^x.$$ Suppose $x=-X$, where $X>1$, then $1+x=1-X<0$ and $e^{x}=e^{-X}=1/e^X>0$. Hence (1) holds.

Now take logarithms of (1) to obtain $$\log(1+x)\leq x.$$ But $$\log(1+x) = x-\frac{x^2}{2}+\frac{x^3}{3}-\cdots,$$ where $|x|<1$. Suppose $x=-X$, where $0<X<1$, then $$\log(1+x)=\log(1-X)=-X-\frac{X^2}{2}-\frac{X^3}{3}-\cdots<-X=x.$$ Or, equivalently, $$1+x<e^x,$$ where $-1<x<0$.

Share Cite Follow edited Sep 25, 2013 at 12:59 answered Sep 25, 2013 at 12:18 pshmath0pshmath0 11k66 gold badges4848 silver badges8888 bronze badges $\endgroup$ 1

• 1 $\begingroup$ Third line, I think you meant "Suppose $x=-X$" and not "Suppose $x<-X$". $\endgroup$ – user37238 Commented Sep 25, 2013 at 12:49

Add a comment | 1 $\begingroup$

The fact $\frac{d}{dx} e^x = e^x$ is nicely demonstrated using the self-similar nature of exponential functions. (See my answer here.)

This justifies (actually, declares) that $y=x+1$ is tangent to $y=e^x$; thereafter, since the slope increases (or decreases) as $x$ gets larger (respectively, smaller) the line and curve cannot meet again (which is an informal way of stating the convexity property).

Share Cite Follow edited Apr 13, 2017 at 12:19 CommunityBot 1 answered Sep 25, 2013 at 23:21 BlueBlue 80.6k1414 gold badges124124 silver badges247247 bronze badges $\endgroup$ Add a comment | 1 $\begingroup$

We know the function $x^x$ has a single local minimum at $x=\frac1e$. Thus, for positive $x$, we have: \begin{align} \left(\frac1e\right)^{1/e}&\le x^x\\ e^{1/e}&\ge\frac1{x^x}\\ e^{1/xe}&\ge\frac1x\\ e^{1/xe-1}&\ge\frac1{xe}\\ e^{(1/xe-1)}&\ge\left(\frac1{xe}-1\right)+1 \end{align} Let $t=\frac1{xe}-1$. If $x>0$, we have $t>-1$. Thus, for all $t>-1$: $$e^t\ge t+1$$ (To prove the above for $t\le -1$, simply note that the left-hand side is always positive while the right-hand side would be zero or negative.) QED.

Share Cite Follow answered Oct 29, 2015 at 1:28 Akiva WeinbergerAkiva Weinberger 24.9k22 gold badges4949 silver badges122122 bronze badges $\endgroup$ Add a comment | 1 $\begingroup$

A proof using only a little basic calculus, and not too many cases:

Set

$\alpha(x) = e^{-x}(1 + x); \tag{1}$

then

$\alpha'(x) = -e^{-x}(1 + x) + e^{-x} = -xe^{-x}, \tag{2}$

and

$\alpha(0) = 1; \tag{3}$

we note that

$x > 0 \Rightarrow \alpha'(x) < 0 \tag{4}$

and

$x < 0 \Rightarrow \alpha'(x) > 0 \tag{5}$

with

$\alpha'(0) = 0; \tag{6}$

thus, for $x > 0$,

$\alpha(x) - 1 = \alpha(x) - \alpha(0) = \int_0^x \alpha'(s) ds < 0, \tag{7}$

whence

$e^{-x}(1 + x) = \alpha(x) < 1; \tag{8}$

when $x < 0$,

$1 - \alpha(x) = \int_x^0 \alpha'(s) ds > 0, \tag{9}$

yielding

$e^{-x}(1 + x) = \alpha(x) < 1 \tag{10}$

in this case as well; (8) and (10) together imply

$1 + x < e^x \tag{11}$

when $x \ne 0$; clearly

$1 + 0 = 1 = e^0; \tag{12}$

combining (11) and (12) shows that

$1 + x \le e^x \tag{13}$

for every $x \in \Bbb R$, with strict inequality precisely when $x \ne 0$. QED.

Share Cite Follow edited Mar 14, 2019 at 0:26 J. W. Tanner 63k44 gold badges4242 silver badges8888 bronze badges answered Aug 19, 2015 at 18:56 Robert LewisRobert Lewis 72.3k55 gold badges6363 silver badges122122 bronze badges $\endgroup$ Add a comment | 1 $\begingroup$

The approximation of the exponential function by its linear Taylor polynomial has remainder term $R(x) := \exp(x) - (1+x)$. The Taylor Remainder Theorem then yields some $\xi$ in between $0$ and $x$ such that $R(x) = \exp''(\xi) x^2 = \exp(\xi) x^2 \ge 0$.

Share Cite Follow answered Nov 4, 2021 at 10:16 community wiki user630227 $\endgroup$ Add a comment | 0 $\begingroup$

This answer uses no calculus or geometry.

Prerequisites: Algebra, basic facts about limits and that for $a \gt 0$ we can define $a^x$ for $x \in \Bbb R$ (see limits of rational exponents).

Theorem 1: There is one and only one number $a \gt 0$ satisfying

$$\tag 1 \forall x \in \Bbb R, \; a^x \ge 1 + x$$

Analyzing $\text{(1)}$, you'll be naturally lead to examine

$\tag 2 u_n \le a \le v_n \text{ where } n \ge 2$ with $\tag 3 u_n = (1 + \frac{1}{n})^n \text{ and } \le v_n = (1 - \frac{1}{n})^{-n}$

Searching, you find answer links from this site:

$\quad u_n \text{ is strictly increasing}:\quad$ here $\quad v_n \text{ is algebraically related to } u_n:\quad$ here $\quad u_n \le 3:\quad$ here

By working with the theory in the above links you will conclude that only one real number, call it $e$, can possibly satisfy $\text{(1)}$. Again, as in the first answer link above, you will use the Bernoulli's inequality and

$\tag 4 e^\frac{s}{t} =\lim_{n\to \infty} \Big( 1+\frac{\frac{s}{t}}{\frac{ns}{t}} \Big)^{\frac{ns}{t}}$

to wrap things up:

$$ \forall x \in \Bbb R, \; e^x \ge 1 + x$$ Share Cite Follow edited Jun 12, 2020 at 10:38 community wiki 2 revsCopyPasteIt $\endgroup$ Add a comment |

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged

• inequality
• exponential-function
• big-list

.

• Featured on Meta
• The December 2024 Community Asks Sprint has been moved to March 2025 (and...
• Stack Overflow Jobs is expanding to more countries

Linked

60 How to prove that $\log(x)<x$ when $x>1$? 7 Prove that $e^x\ge x+1$ for all real $x$ 7 To show that $e^x > 1+x$ for any $x\ne 0$ 5 Showing that $e^x > 1+x$ 3 Proof of $e^x - 1 \geq x$ for ${x: -1 \leq x < 0}$ 0 how to prove that $\exp(x)\geq x+1$ 0 Prove that $e^x>x+1 \forall x\ne 0$ 0 How can we show that $e^{-2\lambda t}\lambda^2\le\frac1{e^2t^2}$ for all $\lambda,t\ge0$? 2 How to prove this inequality for exp? 0 Prove the inequality with the mean value/Lagrange theorem See more linked questions

Related

1 Is most of the GM-AM Inequality in its codicil? 15 Why does the Cauchy-Schwarz inequality hold in any inner product space? 31 Puzzles or short exercises illustrating mathematical problem solving to freshman students 8 Short and intuitive proof that $\left(\frac{n}{k}\right)^k \leq \binom{n}{k}$ 15 which exact integration techniques belong in a first year calculus/analysis course? 18 University-level books focusing on intuition? 11 Inequality for absolute values 0 Interesting real life applications of elementary mathematics 106 What seemingly innocuous results in mathematics require advanced proofs?

Hot Network Questions

• Is it possible to get symbolic integral for this?
• How to accept the completion text in PowerShell terminal?
• The third person preterite of "CRIAR" is "crio" without accent instead of "crió¨with accent?
• How to decompose the following rational function into partial fractions?
• Orly airport Metro ticket information
• Handling One-Inflated Count Data Instead of Zero-inflated
• Kids' book where a girl trades her heart for a heart made of lead
• Colombian passport expires in 5 months
• Mama’s cookies too dry to bake
• What is a good way to DM when the party searches for something?
• What religious significance does the fine tuning argument have?
• Review request: evolution of dragon "fire"
• Free Kei Friday
• Why are so many problems linear and how would one solve nonlinear problems?
• How serving documents ensure that a party got the right ones?
• What is this FreeDOS kernel loader found on the “W3x4NTFS” disk image?
• Why does capacitive coupling require a base resistor in an emitter follower?
• Why did my pancake stick to my pan?
• I saw this article online that mentioned animal testing for immersion in water for applications in astronauts. What plans might require this?
• LGPL-like license for 3D models
• What bladed melee weapon would be best suited for a warrior in zero-gravity?
• Search warrants - do the item(s) being searched for limit the scope of the search?
• Did a peaceful reunification of a separatist state ever happen?
• Should I include my legal name on my CV if my preferred name is not reflected on my diplomas? -- Humanities

more hot questions Question feed Subscribe to RSS Question feed

To subscribe to this RSS feed, copy and paste this URL into your RSS reader.