2
$\begingroup$

I need to prove that $ -2 + x + (2+x)e^{-x}>0 \quad \forall x>0$

If I define $f(x) = -2 + x + (2+x)e^{-x}$ and plot it I can see it's a monotonously growing function, and f(0)=0. Then $f(x)>0$ if $x>0$.

However I can't find the way to prove this. Ideally I would like to prove it without deriving the function and by using some inequalities, however I don't know if it is possible. Any hint is really appreciated.

$\endgroup$
1
2
$\begingroup$

We have to prove that $$e^{-x}>{2-x\over 2+x}\qquad(x>0)\ .\tag{1}$$ This is obvious when $x\geq2$. For $0<x<2$ replace the claim $(1)$ by $$e^x<{2+x\over2-x}=1+{x\over 1-{x\over2}}\qquad(0<x<2)\ .$$This means that we have to prove $$1+x+{x^2\over2!}+{x^3\over3!}+{x^4\over4!}+\ldots<1+x+{x^2\over2}+{x^3\over2^2}+{x^4\over 2^3}+\ldots\qquad(0<x<2) ,$$ which follows immediately by comparing terms.

$\endgroup$
1
  • $\begingroup$ Just to clarify, although some other answers seems to valid as well, I found this the most elegant and simple. Thus, I am going to accept this one as the correct one. $\endgroup$
    – Iván
    Oct 26 '18 at 17:06
4
$\begingroup$

We have that $f(0)=0$ and

$$f(x)=-2 + x + (2+x)e^{-x}\implies f'(x)=e^{-x}(e^x-1-x)$$

As alternative note that for $x\ge2$ the inequality is trivially satisfied then consider $0<x<2$ and we have

$$-2 + x + (2+x)e^{-x}>0 \iff e^x<\frac{2+x}{2-x}$$

and by $x=\frac2y$ with $y>1$ we obtain

$$f(y)=\left(\frac{y+1}{y-1}\right)^y=\left(1+\frac{2}{y-1}\right)^y>e^2$$

which, for my knowledge, can’t be easily proved for $y\in\mathbb{R}$ without derivatives, namely showing that $f(y)$ is monotonic.

The monotonicity of $f(y)$ can be easily proved for $y\in\mathbb{N}$ and extended to the real case assuming that $f(y)$ is convex. We could also try to extend the result to reals through rationals.

$\endgroup$
6
  • $\begingroup$ <<Ideally I would like to prove it without deriving the function and by using some inequalities...>> $\endgroup$
    – rtybase
    Oct 23 '18 at 21:22
  • $\begingroup$ @rtybase Ideally means solely? I thought that it could be also useful give a solution in that classic way. $\endgroup$
    – user
    Oct 23 '18 at 21:25
  • $\begingroup$ @gimusi Are you sure that the inequality holds $\forall x>0$? I think that for $0<x<2$ it is not true. $\endgroup$
    – Iván
    Oct 23 '18 at 22:47
  • $\begingroup$ @Iván Yes indeed but for $x>2$ the original inequality is sum of positive terms. $\endgroup$
    – user
    Oct 24 '18 at 2:38
  • $\begingroup$ @gimusi sorry to insist, but I plotted the expression that you proposed and it is not true that the inequality holds. May be I am missing something, here you have a plot to check it: ibb.co/nL2BmV. I understand that for $x>2$ $f(x)>0$, but the range $0<x<2$ is the hard part to prove, and the expression that you provided (thank you by the way) I think that does not hold in that range. $\endgroup$
    – Iván
    Oct 24 '18 at 17:03
3
$\begingroup$

$$ \frac{d^2}{dx^2}\left[-2+x+(2+x)e^{-x}\right] = x e^{-x} $$ hence $f(x)=-2+x+(2+x)e^{-x}$ is a convex function on $\mathbb{R}^+$.
Since $f'(0)=f(0)=0$, $f(x)$ is increasing and positive on $\mathbb{R}^+$.

$\endgroup$
2
$\begingroup$

This approach requires an extra condition $x\lt2.$

\begin{align}\ln\left(1+\dfrac{x}{2}\right)-\ln\left(1-\dfrac{x}{2}\right)&=\left[\dfrac{x}{2}-\dfrac{1}{2}\left(\dfrac{x}{2}\right)^2+\dfrac{1}{3}\left(\dfrac{x}{2}\right)^3+\cdots\right]-\left[-\dfrac{x}{2}-\dfrac{1}{2}\left(\dfrac{x}{2}\right)^2-\dfrac{1}{3}\left(\dfrac{x}{2}\right)^3-\cdots\right]\\&=x+\dfrac{2}{3}\left(\dfrac{x}{2}\right)^3+\cdots\\&\gt x\quad\forall x\lt2\cdots(*)\\\end{align}

$\forall x\gt0,\dfrac{2+x}{2-x}\gt1\implies\ln\left(\dfrac{2+x}{2-x}\right)\gt0$

$\therefore0\lt x\lt\ln\left(\dfrac{2+x}{2-x}\right)\implies e^{-x}\gt\dfrac{2-x}{2+x}\implies -2+x+(2+x)e^{-x}\gt0$

$\endgroup$
1
  • $\begingroup$ That's also a nice way, of course you are assuming $0<x<2$ but maybe it should be specified. I was also trying to find a method which doesn't make use of derivative, convexity, integrals or series concept but only algebraic methods. $\endgroup$
    – user
    Oct 25 '18 at 10:37
2
$\begingroup$

A variation on Christian Blatter's solution:

$e^{-x} (2+x) +x-2 = e^{-x} ( 1+ \frac{x-2}{x+2}) $

As $e^{-x} \gt \ 0 \ \forall \ x \in \mathbb{R} $, it's enough to check if, $(1+\frac{x-2}{x+2}) \gt 0 $ for $x\gt 0$

However, now we have: $1+\frac{x-2}{x+2} = 1+\frac{x-2+2-2}{x+2} = 2-\frac{4}{x+2}$ which is strictly increasing for $x\gt 0$ and bounded below by $ 0 = e^{-x} ( 1+ \frac{x-2}{x+2}$ ) for $x=0$

$\endgroup$
0
$\begingroup$

Another alternative that I found but is not as elegant as the answer marked as the accepted one is expanding the exponential and rearranging the terms as follows:

$ \left(e^{x } (x -2 )+2 +x \right) >0 \\ \iff (1+x+\frac{x^2}{2!}++\frac{x^3}{3!}+\dots) (x -2 )+2 +x >0 \\ \iff \sum_{n=3}^{\infty} x^n(\frac{1}{(n-1)!}-\frac{2}{(n-1)!n})>0 $

$\endgroup$

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.