46

Show that $U_n:=\left(1+\dfrac{1}{n}\right)^n$, $n\in\Bbb N$, defines a monotonically increasing sequence.

I must show that $U_{n+1}-U_n\geq0$, i.e. $$\left(1+\dfrac{1}{n+1}\right)^{n+1}-\left(1+\dfrac{1}{n}\right)^n\geq0.$$

I am trying to go ahead of this step.

Mars Plastic
  • 4,239
  • 1
    See also http://math.stackexchange.com/questions/64860/proving-bigl1-frac1n1-bigrn1-gt-1-frac1nn –  Mar 08 '14 at 19:36

14 Answers14

67

We use the inequality between the geometric mean and the arithmetic mean for the following positive numbers $$ x_{1}=1,~x_{2}=x_{3}=\ldots=x_{n+1}=1+\frac{1}{n}\text{.}% $$ Then $$ \sqrt[n+1]{x_{1}x_{2}\cdots x_{n+1}}<\frac{x_{1}+x_{2}+\ldots+x_{n+1}}{n+1}% $$ (the inequality is strict, since the numbers can't be all equal) translates to $$ \left( 1+\frac{1}{n}\right) ^{\frac{n}{n+1}}<\frac{1+n\left( 1+\frac{1}{n}\right) }{n+1}=1+\frac{1}{n+1}% $$ hence $a_{n}<a_{n+1}$.

digital-Ink
  • 1,986
40

Let $x_n=\big(1+\frac{1}{n}\big)^n$ be the $n^\text{th}$ element of the sequence. Then $$ \begin{align} \frac{x_{n+1}}{x_{n}} &=\frac{(1+\frac{1}{n+1})^{n+1}}{(1+\frac{1}{n})^{n}} \\&=\bigg(\frac{1+\frac{1}{n+1}}{1+\frac{1}{n}}\bigg)^n\bigg(1+\frac{1}{n+1}\bigg) \\&=\bigg(1-\frac{1}{(n+1)^2}\bigg)^n\bigg(1+\frac{1}{n+1}\bigg) \\&≥^*\bigg(1-\frac{n}{(n+1)^2}\bigg)\bigg(1+\frac{1}{n+1}\bigg) \\&>\bigg(1-\frac{n+1}{(n+1)^2}\bigg)\bigg(1+\frac{1}{n+1}\bigg) \\&=1 \end{align}$$ The $\ge^*$ step follows from Bernoulli's inequality, which says that $(1+x)^m\ge 1+mx $ whenever $m\in \mathbb N$ and $x\ge -1$.

Mike Earnest
  • 75,930
Mikasa
  • 67,374
  • @Olivier: Thanks for editing. – Mikasa Jul 07 '12 at 16:17
  • 2
    no problem, to make larger paranthesis, use the command \big(,\big) or \bigg(,\bigg) for extra large ones. You can also use the command \left(,\right) and LateX will automatically scale the size so that it looks nice. – Olivier Bégassat Jul 07 '12 at 17:56
  • 6
    I think the inequality $$\left(1-\frac{1}{(n+1)^2}\right)^n\geq 1-\frac{n}{(n+1)^2}$$should be justified by at least mentioning Bernoulli's inequality: $,\forall -1<x\in \Bbb R,,,,(1+x)^n\geq 1+nx,$. Anyway, this proof is more elementary. +1 – DonAntonio Jul 08 '12 at 02:26
  • 1
    @DonAntonio: Yes. This proof is more elementary as you noted. Honestly, I knew it from the book Modern Calculus and Analytic Geometry by R. A. Silverman. Thanks – Mikasa Jul 08 '12 at 05:10
  • 2
    @Babak Sorough Thanks. I have chosen this answer. $$Will you please explain$$ $$How\bigg(1−\frac{n}{(n+1)^2}\bigg)\geq\bigg(\frac{1}{1+\frac{1}{n+1}}\bigg)?$$ – rajendra bakre Jul 11 '12 at 08:41
  • @rajendrabakre: Thanks for choosing mine, but note that other answers like Olivier's will give you more insight about the question. I edited the answer. :) – Mikasa Jul 11 '12 at 11:48
  • ++Helpful, once again! :^D – amWhy Mar 08 '13 at 01:44
  • @amWhy: Thanks Amy ;-) – Mikasa Mar 08 '13 at 06:13
  • 29
    The analysis flows more easily if we write $$\frac{\left(1+\frac1{n+1}\right)^{n+1}}{\left(1+\frac1n\right)^n}=\left(1+\frac1n\right),\left(1-\frac{1}{(n+1)^2}\right)^{n+1}\ge \left(1+\frac1n\right),\left(1-\frac{1}{n+1}\right)=1$$Doesn't that seem easier? ;-)) -Mark – Mark Viola Aug 20 '16 at 22:17
27

The (well known) elementary proof that this sequence is increasing relies on the Bernoulli inequality, which states that, for real $x\ge -1$ and $n\in \mathbb{N}$, $$(1+x)^n \ge 1+nx$$ which can be easily shown by induction. This looks quite inefficient but should not be underestimated. If you know this, then observe that $$\left(1+\frac{1}{n}\right)^{n} > \left(1+\frac{1}{n-1}\right)^{n-1} $$ is equivalent to $$\left( \frac{1+\frac{1}{n}}{1+\frac{1}{n-1}}\right)^{n} > \left(1+\frac{1}{n-1}\right)^{-1} = 1-\frac{1}{n}$$ The lhs is equal to $$ \left(\frac{n^2-1}{n^2}\right)^n = \left(1-\frac{1}{n^2}\right)^n $$ which, according to Bernoulli is $$> 1-\frac{n}{n^2} = 1-\frac{1}{n}$$ which is what was to be shown.

Thomas
  • 22,361
14

Take logarithms. You need to compare $n\ln(1+\frac{1}{n})$ to $(n+1)\ln(1+\frac{1}{n+1})$. Because the logarithm is strictly concave, the function (defined for positive $x$) $$\frac{\ln(1+x)}{x}=\frac{\ln(1+x)-\ln(1)}{(1+x)-1}$$ is strictly decreasing (and tends to $1=\ln'(1)$ as $x$ tends to $0$.) Apply this to the striclty decreasing sequence $1/n$ and you get that the sequence $$\frac{\ln(1+1/n)}{1/n}\mathrm{~is~strictly~increasing.}$$ Of course $\frac{\ln(1+1/n)}{1/n}=n\ln(1+\frac{1}{n})$, so, upon exponentiating, $U_n$ is strictly increasing (and tends to $e$.)

10

If you expand $(1+\frac1n)^n$ by the binomial theorem, the term involving $1^{n-k}(\frac1n)^k$ is $\binom{n}{k}/n^k$ (I take such a term to exist, and be $0$, in case $k>n$). If one can show that each such term is a monotonically increasing expresion in $n$, then certainly the sum of all terms will be a monotonically increasing expression in $n$ (this involves formally adding up infinitely many expressions, but in comparing $U_n$ and $U_{n+1}$ only finitely many terms are involved, so there is no need to take limits). Now we can write $$ \frac{\binom{n}{k}}{n^k}=\frac1{k!}\cdot\frac{n}{n}\frac{(n-1)}n\cdots\frac{(n-k+1)}n =\frac1{k!}(1-\frac1n)(1-\frac2n)\ldots(1-\frac{(k-1)}n) $$ This expression is zero as long as $n<k$, and beyond that point all factors are positive and either independent of $n$ or increasing expressions in $n$. We may conclude that term $k$ is constant for $k\leq1$, and a weakly increasing function of $n$, strictly increasing as soon as it is nonzero, for $k\geq2$. This proves the result.

5

My two cents (as later similar question closed):

$$\frac{a_{n+1}}{a_n}=\frac{\left( 1+ \frac{1}{n+1} \right)^{n+1}}{\left( 1+ \frac{1}{n} \right)^{n}}=\frac{n+1}{n}\left( 1- \frac{1}{(n+1)^2} \right)^{n+1}>\\ >\frac{n+1}{n}\left( 1- \frac{1}{(n+1)} \right)=1$$

zkutch
  • 13,410
  • I like this answer best. It's the simplest. On the other hand, from the accepted answer I learned that $(n+2)(n^2+n+1)≥(n+1)^3, $ which I wasn't previously aware of. – Adam Rubinson Jul 12 '21 at 10:56
4

Wanted to add yet another method to the "catalogue"; I learned this by solving a problem from the book Numbers and Functions: Steps into Analysis by R P. Burn.

First, we have the following lemma:

Let $0 < a < b$. Then $$\frac{b^{n+1}-a^{n+1}}{b-a} < (n+1)b^n$$

This can be proved by noting that $$b^{n+1}-a^{n+1} = (b-a)(b^n + b^{n-1} a + b^{n-2} a^2 + \dots + a^n)$$ and since $0 < a < b \implies 0 < a^n < b^n$, we get $$b^n + b^{n-1} a + b^{n-2} a^2 + \dots + a^n < b^n + b^{n-1} b + b^{n-2} b^2 + \dots + b^n = (n+1)b^n$$.

Now plug in $$a = 1 + \frac{1}{n+1}, \hspace{10pt} b = 1 + \frac{1}{n}$$ into the lemma, and let $$t_n = \left(1 + \frac{1}{n}\right)^n$$ This gives

$$\frac{\overbrace{\left(1+\frac{1}{n}\right)^{n+1}}^{(1+\frac{1}{n})t_n} - \overbrace{\left(1 + \frac{1}{n+1}\right)^{n+1}}^{t_{n+1}}}{\frac{1}{n}-\frac{1}{n+1}} < (n+1)\overbrace{\left(1+\frac{1}{n}\right)^n}^{t_n}$$ which after some straightforward algebraic simplification yields $t_n < t_{n+1}$.

Aky
  • 607
  • I saw a similar proof in a book I was reading. Can you give me an idea on how can one come up with something so elaborate? – eem Jul 26 '19 at 17:26
3

HINT: Differentiate with respect to $n$. Prove this is always increasing.

Side note: Do you know what $a_n$ is?


So we have $f(x) = \left(1+\frac1{n}\right)^n$

$\log(f(x)) = n\log\left(1+\frac1{n}\right)$

$\log(f(x)) = n\log\left(n+1\right) - n\log(n)$

Try differentiating now.

Guy
  • 8,857
  • 1
  • 28
  • 57
3

Note that the function $f(x) = (1+\frac 1x)^x$ is differentiable. We may find its derivative as follows: $$ \ln(f(x)) = \ln\left((1+\frac 1x)^x\right) = x\ln(1 + \frac 1x)\\ f'(x)/f(x) = \ln\left(1 + \frac 1x\right) + \frac{x}{1+\frac 1x}\cdot \frac{-1}{x^2}\\ f'(x)/f(x) = \ln\left(1 + \frac 1x\right) - \frac{1}{x+1}\\ f'(x) = \left[\ln\left(1 + \frac 1x\right) - \frac{1}{x+1}\right]f(x)\\ f'(x) = \left[\ln\left(\frac {x+1}x\right) - \frac{1}{x+1}\right]f(x)\\ f'(x) = \left[\ln(x+1) - (\ln(x) + \frac{1}{x+1})\right]f(x) $$ Show that the derivative is positive from $x = 2013$ to $x = 2014$.

Ben Grossmann
  • 225,327
  • 4
    What is $\ln$? It is the logarithm with base $\mathrm{e}$. But who is $\mathrm{e}$? Well, it is the limit of $a_n$. But why is $a_n$ convergent? Because it is increasing and bounded. Do you see the chicken and the egg in your approach? – digital-Ink Mar 08 '14 at 19:14
  • 2
    @digital-Ink I define $\ln x$ to be the function whose derivative is $1/x$ and whose value at $x = 1$ is $0$. – Ben Grossmann Mar 08 '14 at 19:15
  • @digital-Ink $e$ has multiple definitions, and though historically your argument would be correct, in this context we can use it. Although you will most definitely get my upvote if you do something without it. – Guy Mar 08 '14 at 19:15
  • 1
    @digital-Ink I define $e$ to be the unique positive number $a$ such that the area under the curve $a^x$ from $a$ to $1$ is $(a-1)$. Oh my. It comes out be equal to this series too? What a coincidence. – Guy Mar 08 '14 at 19:17
  • @omnomnomnom how can you say $\ln\left(1 + \frac 1x\right) \geq \frac{1}{x+1}$ – raj Mar 08 '14 at 19:19
  • It's a little easier to deal with in the form I have it in now – Ben Grossmann Mar 08 '14 at 19:24
  • @Sabyasachi: I remember well that in high school we used just algebraic properties (like the binomial expansion formula) to prove that $a_n$ is increasing and bounded (actually, we used it together with $b_n=(1+\frac{1}{n})^{n+1})$, then we introduced $\mathrm{e}$ as the limit of these sequences. This seems the most elementary. – digital-Ink Mar 08 '14 at 19:24
  • @Sabyasachi: Clearly, the path to the truth are many. But it seems wrong to go around through all the Calculus (derivatives, anti-derivatives, integrals or series), just to prove a fundamental property. At least, that's my view. – digital-Ink Mar 08 '14 at 19:27
  • @digital-Ink yes "elementary" yes. fun? no. – Guy Mar 08 '14 at 19:28
  • @digital-Ink clearly yes and I agree. but for some reason I view calculus as completely fundamental, as fundamental as arithmetic. Yes, it took us a lot of time to develop calculus, but once you have understood it, isn't the beauty immediately apparent? Binomial theorem will get messy here, although in some cases that IS applied elegantly, it might be better than the application of calculus here. Feel free to post something like that if you have it. Cheers. :) – Guy Mar 08 '14 at 19:31
  • @Sabyasachi: I will try to write an algebraic proof. Unless someone else writes it first. – digital-Ink Mar 08 '14 at 19:34
  • @ Omnomnomnom we can take $ g(x) = \ln(x+1) - (\ln(x) + \frac{1}{x+1} ) $ and prove that g(x) is increasing – raj Mar 08 '14 at 19:34
  • @raj I suppose so; all we really need to show is that it is always positive. See the link I've provided at the end of my answer for an alternate approach. – Ben Grossmann Mar 08 '14 at 19:35
  • @digital-Ink that'd be great. variety is the spice of life. :D – Guy Mar 08 '14 at 19:39
  • @Sabyasachi: Done. No need for binomial expansion. – digital-Ink Mar 08 '14 at 19:53
  • @Sabyasachi: Also, I assure you that there is no need for calculus in order to prove the inequality between the geometric mean and the arithmetic mean (of course, using the logarithm again as a concave function trough Jensen's inequality would be elegant) – digital-Ink Mar 08 '14 at 19:59
  • @digital-Ink yes I am aware of the fact that there is no need of calculus to prove $AM \ge GM$. I lost you though, where does this come in here? – Guy Mar 08 '14 at 20:02
  • oh. point in sarcasm? Got it. – Guy Mar 08 '14 at 20:02
  • @Sabyasachi: I used the inequality in order to prove that $a_n$ is increasing. Please, see below. No sarcasm included! Just wanted to point out that the inequality between means can be considered an algebraic tool. – digital-Ink Mar 08 '14 at 20:22
  • Then link at the end of this answer is one to the very question above. Is this some kind of humour that I am missing? – Marc van Leeuwen Apr 08 '14 at 11:05
  • @MarcvanLeeuwen I... would you believe I did this by accident? I think there was another question I had meant to link to, but honestly I can't remember. Thank you for pointing this out. – Ben Grossmann Apr 08 '14 at 14:28
3

Let $u_n = (1+1/n)^n$ and $v_n = (1+1/n)^{n+1}$. Then for $n>1$

\begin{align}\frac{u_n}{v_{n-1}}=\bigg(\frac{n+1}{n}\bigg)^n\bigg(\frac{n-1}{n}\bigg)^n =\bigg(1-\frac{1}{n^2}\bigg)^n\\ > 1-\frac{1}{n}=\frac{n-1}{n} \end{align}

So \begin{align}u_n> v_{n-1}\bigg(\frac{n-1}{n}\bigg)=\bigg(1+\frac{1}{n-1}\bigg)^{n-1}=u_{n-1}\end{align}

Jose Antonio
  • 7,154
2

We show that $$ \left(1+\frac{1}{n-1}\right)^{n-1} \leq \left(1+\frac{1}{n}\right)^{n} $$ for any $n \geq 2$.

For any $n \geq 2$ we have that $$ n \int^n_{n-1} \frac{1}{x(x+1)}dx \leq \int^n_{n-1} \frac{1}{x}dx $$ because $\frac{n}{x+1} \leq 1$ for all $x \in [n-1,n]$. But this inequality is equivalent to $$ n \int^n_{n-1} \frac{1}{x}-\frac{1}{x+1}dx \leq \int^n_{n-1} \frac{1}{x}dx \iff\\ \iff (n-1)\int^n_{n-1} \frac{1}{x}dx \leq n\int^n_{n-1} \frac{1}{x+1}dx \iff\\ \iff (n-1)\int^n_{n-1} \frac{1}{x}dx \leq n\int^{n+1}_{n} \frac{1}{x}dx \iff\\ \iff (n-1)\ln\left(\frac{n}{n-1}\right) \leq n \ln\left(\frac{n+1}{n}\right) \iff\\ \iff \left(1+\frac{1}{n-1}\right)^{n-1} \leq \left(1+\frac{1}{n}\right)^{n}. $$

2

We have that

$$\left(1+\dfrac{1}{n+1}\right)^{n+1}-\left(1+\dfrac{1}{n}\right)^n>0 \iff \ln \left(1+\frac{1}{n+1}\right)^{(n+1)} - \ln \left(1+\frac{1}{n}\right)^n>0$$

and

$$(n+1) \ln \left(1+\frac{1}{n+1}\right) - n \ln \left(1+\frac{1}{n}\right) \ge(n+1)\left[ \ln \left(1+\frac{1}{n+1}\right) - \ln \left(1+\frac{1}{n}\right)\right]=$$

$$=(n+1)\left[ \ln \left(\frac{n+2}{n+1}\right) - \ln \left(\frac{n+1}{n}\right)\right]=$$

$$=(n+1) \ln \left(\frac{(n+2)(n+1)}{(n+1)^2}\right)=(n+1) \ln \left(\frac{n^2+2n+2}{n^2+2n+1}\right) >0$$

since $\log x>0$ for $x>1$.

user
  • 154,566
2

By Lagrange's Mean Value Theorem.

$$ \exists\xi\in(x,x+1):\frac{\ln(x+1)-\ln x}{(x+1)-x}=\frac{1}{\xi}>\frac{1}{x+1} $$

Thus,

$$ \frac{d}{dx}\ln\left(1+\frac1x\right)^x=\ln(x+1)-\ln x-\frac{1}{x+1}>0 $$

This shows that $\left(1+\dfrac1x\right)^x$ is monotonically increasing.

O-17
  • 438
0

Here's a proof I learned by solving problem 90 from Section 11.1 of James Stewart's Calculus: Early Transcendentals (8th edition). We'll need the following result:

Lemma: If $0\leq a<b$ and $n$ is a positive integer, then $$\frac{b^{n+1}-a^{n+1}}{b-a}<(n+1)b^n$$ Proof: Define the function $f:[0,\infty)\rightarrow\mathbb{R}$ by $f(x)=x^{n+1}$, where $n$ is a positive integer. Then $f$ is differentiable (and hence continuous) over $[0,\infty)$. Moreover, by the power rule, $f'(x)=(n+1)x^n$ for all $x\geq 0$, which is strictly increasing because $n$ is a positive integer (this will be important later, so we'll tuck it in our back pocket).
Since $f$ is differentiable over $[0,\infty)$, and $0\leq a<b$, it is necessarily continuous over $[a,b]$ and differentiable over $(a,b)$, so the mean value theorem asserts that there exists $c\in(a,b)$ such that $$f'(c)=\frac{f(b)-f(a)}{b-a} =\frac{b^{n+1}-a^{n+1}}{b-a}$$ Since $f'$ is strictly increasing and $c<b$, it follows that $f'(c)<f'(b)=(n+1)b^n$, so $$\frac{b^{n+1}-a^{n+1}}{b-a}=f'(c)<f'(b)=(n+1)b^n$$ It immediately follows that $\frac{b^{n+1}-a^{n+1}}{b-a}<(n+1)b^n$.

Simple algebraic manipulation shows that $\frac{b^{n+1}-a^{n+1}}{b-a}<(n+1)b^n$ is equivalent to $b^n[(n+1)a-nb]<a^{n+1}$. With this established, we choose $a=1+\frac{1}{n+1}$ and $b=1+\frac{1}{n}$ (this is justified since $\frac{1}{n+1}<\frac{1}{n}$ for all $n>0$, and hence $1+\frac{1}{n+1}<1+\frac{1}{n}$), which gives $$\left(1+\frac{1}{n}\right)^n\left[(n+1)\left(1+\frac{1}{n+1}\right)-n\left(1+\frac{1}{n}\right)\right]<\left(1+\frac{1}{n+1}\right)^{n+1}$$ Therefore, $$\left(1+\frac{1}{n}\right)^n(n+1+1-n-1)<\left(1+\frac{1}{n+1}\right)^{n+1}$$ and thus, $$\left(1+\frac{1}{n}\right)^n(1)=\left(1+\frac{1}{n}\right)^n<\left(1+\frac{1}{n+1}\right)^{n+1}$$ This shows that $U_n=\left(1+\frac{1}{n}\right)^n$ is strictly increasing.

Alann Rosas
  • 5,421
  • 1
  • 9
  • 28