Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I don't quite remember where this problem is from. I came across is sometime last summer, when I was in an olympiad-problem mood and I decided to improve my inequality skills.

Suppose $a,b,c > 0$. Then we want to show that

$$\frac{16}{27} \left( \frac{a}{b+c} + \frac{b}{a+c} +\frac{c}{a+b} \right) ^3 + \left( \frac{abc}{(a+b)(b+c)(a+c)}\right)^{1/3} \geq \frac{5}{2}$$

I think that there are many things to notice. Firstly, it's homogenous. The left part is tantalizingly close to Nesbitt's inequality. The right part seems to demand AM-GM attention.

share|cite|improve this question
I think the right part has to be $\frac{(a+b)(b+c)(a+b)}{abc}$not $\frac{abc}{(a+b)(b+c)(a+b)} $ – Babak Miraftab Jan 31 '12 at 7:03
And I think it should be $\frac{abc}{ \textbf{(a+c)}(b+c)(a+b)}$. – Gigili Jan 31 '12 at 8:38
Shouldn't the factors of the denominator of the right-hand term be the denominators of the left-hand term: namely $b+c$, $c+a$, and $a+b$ ? – John Bentin Jan 31 '12 at 9:27
I would like to have a "certified" version of the inequality. Indeed the left term is greater or equal to 2 by Nesbitt, but the right term is lower or equal than $frac{1}{2}$ instead of being greater or equal. On the other hand I didn´t find a counterexample, so I think that the two terms "interact" for giving the inequality. On the right hand side the denominator should be as John pointed out, this is why I would like the confirmation it is the only misprint there. – Giovanni De Gaetano Jan 31 '12 at 14:40
Whoops! Yes, I'm sorry about the $a + b$ twice! – mixedmath Jan 31 '12 at 15:12
up vote 11 down vote accepted

First, make the substitutions $$ x= \frac{a}{b+c}, \quad y= \frac{b}{a+c}, \quad z= \frac{c}{a+b}. $$ The strategy will be to reduce the problem to an inequality in the single variable $t=(xyz)^{1/3}$. Note that $xy+yz+xz+2xyz=1$, and the inequality to be proved is $$ \frac{16}{27}\left(x+y+z\right)^3+(xyz)^{1/3}\geq\frac{5}{2}. $$ Now $$ \frac{(x+y+z)^2}{3}\geq xy+yz+xz=1-2xyz $$ and also $x+y+z\geq3/2$ by Nesbitt's inequality. Therefore, $$ \frac{16}{27}(x+y+z)^3=\frac{16}{9}\cdot(x+y+z)\cdot\frac{(x+y+z)^2}{3}\geq\frac{8}{3}(1-2xyz), $$ and it is sufficient to prove the inequality $$ \frac{8}{3}(1-2xyz)+(xyz)^{1/3}\geq\frac{5}{2}. $$ Now $xyz\leq1/8$, because AM-GM gives $8abc\leq (a+b)(b+c)(a+c)$ by grouping pairs on the right-hand side (e.g., $2abc\leq a^2b+bc^2$). Thus by setting $t=(xyz)^{1/3}$, we are reduced to proving that the polynomial $$ f(t):= 8\left(\frac{1-2t^3}{3}\right)+t-\frac{5}{2} = \frac{1}{6}+t-\frac{16}{3}t^3. $$ is nonnegative for $t\in[0,1/2]$. Since $f(0)>0$ and $f(1/2)=0$, we can show that $f(c)>0$ whenever $c$ is a critical point of $f$. But $f'(t)=1-16t^2$, which has $c=1/4$ as its only zero in $[0,1/2]$. As $f(1/4)=4/12>0$, we are done.

share|cite|improve this answer
Nice! That construction is very clean (or as clean as these things get!) – Isaac Solomon Feb 1 '12 at 2:24

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.