Show that for $x$, $y$ real numbers, $0<x$ , $0<y$

$$\left(\frac{x^2 + y^2}{4}\right) < e^{x+y-2}. $$

Someone can help me with this please...

  • $\begingroup$ Probably some other conditions, get into trouble with negatives. $\endgroup$ – André Nicolas Jun 30 '13 at 14:51
  • $\begingroup$ yea only with x,y strictly positive or zero $\endgroup$ – Knight Jun 30 '13 at 14:52

From the well known inequality: $1+z\leq e^z$, we replace $z$ by $\frac{z}{2}-1$ to get $\frac{z}{2}\leq e^{\frac{z}{2}-1}$. Square both sides to get $\frac{z^2}{4}\leq e^{z-2}$. Now let $z=x+y$, to get: $\frac{(x+y)^2}{4}\leq e^{x+y-2}$. Hence:

$$\frac{x^2}{4}+\frac{y^2}{4}<\frac{(x+y)^2}{4}\leq e^{x+y-2}$$


$\frac{x^2}{4}+\frac{y^2}{4}< \frac{(x+y)^2}{4}$ because $x,y> 0$

  • $\begingroup$ @MhenniBenghorbal Thank you ! $\endgroup$ – Amr Jun 30 '13 at 15:16
  • $\begingroup$ @MhenniBenghorbal How about my answer here math.stackexchange.com/questions/432999/… $\endgroup$ – Amr Jun 30 '13 at 15:39
  • $\begingroup$ yea thanxs @Arm (+1) $\endgroup$ – Knight Jun 30 '13 at 16:26
  • $\begingroup$ @Dann Knight you are welcome ! My name is Amr :) $\endgroup$ – Amr Jun 30 '13 at 16:52
  • $\begingroup$ Amr yea sorry! can you help with this?math.stackexchange.com/users/84186/knight $\endgroup$ – Knight Jun 30 '13 at 19:44

When $x+y$ is fixed $x^2+y^2=(x+y)^2-2xy$ takes maximal value when $xy=0.$ So it is enough to prove our inequality when one variable is $0.$ In this case, it can be reduced to $\frac{x^2}{4}\le e^{x-2}$ or $e^z\ge 1+z+\frac{z^2}{4}$ which immediately follows from the fact that $e^z\ge 1+z+\frac{z^2}{2}.$

Alternative way to present solution: Let $x+y=t$ $$\frac{x^2+y^2}{4}=\frac{(x+y)^2-2xy}{4}\le \frac{t^2}{4}\le e^{t-2}$$ and the rest is the same as above.

  • $\begingroup$ I do not understand when you said takes maximal value when xy=0 $\endgroup$ – Knight Jun 30 '13 at 15:08
  • $\begingroup$ I have included another way to explain this fact. $\endgroup$ – leshik Jun 30 '13 at 15:13
  • $\begingroup$ I dont see why or how you can reduce the case to your inequality for $e^y$ - surely if we are setting either variable as being zero we would just get the $e^{x-2}$ inequality again? $\endgroup$ – Andrew D Jun 30 '13 at 15:14
  • $\begingroup$ take $y=x-2$ and expand. Is should have probably used another variable instead of $y.$ $\endgroup$ – leshik Jun 30 '13 at 15:18

This is only a partial answer because it works only for $x+y>2$, but it might give you some insight:

First , we need to reduce this to one variable, so let's use the inequality $\sqrt{a+b}<\sqrt{a}+\sqrt{b}$ (this one is easily proved). We take the square root of both sides, and then apply this inequality to the LHS:


Let $r=x+y$ , and let's rewrite this:

$\frac r2 <{e^\frac r2 -1}$ (I don't know why it's not showing it but the $-1$ is in the exponent from now on)

If we take the derivative of both sides with respect to $r$ variables, we see that we need to prove that

$\frac 12 < \frac 12 {e^\frac r2 -1}$ , or $1<{e^\frac r2 -1}$

I'm sure you already know that exponential functions are monotonically increasing, so we don't need to go over that. If you solve ${e^\frac r2 -1}>1$ , you get $r>2$


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.