5

Let $a,b,c>0$ and such $a+b+c=1$ show that $$\dfrac{1}{1-a}+\dfrac{1}{1-b}+\dfrac{1}{1-c}\ge \dfrac{1}{ab+bc+ac}+\dfrac{1}{2(a^2+b^2+c^2)}$$ Let $p=a+b+c=1,ab+bc+ac=q,abc=r$ $$\Longleftrightarrow -4q^3+q^2-3qr+2r\ge 0$$ it seem hard to prove.

why I say it hard prove: use Schur inequality $$p^3-4pq+9r\ge 0\Longrightarrow r\ge\dfrac{4q-1}{9}$$ it remains to prove that $$\dfrac{4q-1}{9}(2-3q)+q^2-4q^3\ge 0$$ In fact, this inequality can't hold (4q-1)

math110
  • 93,304

1 Answers1

1

Let $a+b+c=3u$, $ab+ac+bc=3v^2$ and $abc=w^3$.

Hence, our inequality it's $\frac{9u^2+3v^2}{9uv^2-w^3}\geq3u\left(\frac{1}{3v^2}+\frac{1}{18u^2-12v^2}\right)$,

which is equivalent to $f(w^3)\geq0$, where $f$ is a linear function.

But the linear function gets a minimal value for an extremal value of $w^3$.

$a$, $b$ and $c$ are positive roots of the equation $(x-a)(x-b)(x-c)=0$ or $w^3=x^3-3ux^2+3v^2x$ and we see that the line $y=w^3$ and graph of $y=x^3-3ux^2+3v^2x$ have three common points (draw it!).

Thus, an extremal value of $w^3$ we get for equality case of two variables

and we need to check also the case $w^3\rightarrow0^+$.

  1. $b=c$. After homogenization we can assume $b=c=1$, which gives $a(a-1)^2\geq0$;

  2. $w^3\rightarrow0^+$.

Let $c\rightarrow0^+$. After homogenization we can assume $b=1$,

which gives $(a-1)^2\geq0$. Done!

  • While I do not prefer this method for proofs as it is it encourages too much "handwaving" as style, if one digs deeper it is perfectly valid. Cant see why someone would go to the extent of downvoting this. +1 – Macavity Jun 29 '16 at 04:42
  • @Macavity 1 where do you see "handwaving"? I am ready to explain all step for you. – Michael Rozenberg Jun 29 '16 at 06:28
  • The "draw it" argument is to me, especially when applied to functions with more than two variables, handwaving - it encourages so called "proofs" like the one here http://math.stackexchange.com/questions/1805719/prove-sqrt2xy-sqrt3yz-sqrt4zx-4/1824912#1824912 I am familiar with the $u, v, w$ method and why it works, but if I were not, I would not be so accepting of graphing to show that $w^3$ gets extreme iff when two of the variables are equal or one zero. – Macavity Jun 29 '16 at 06:36
  • Dear @Macavity ! We need to draw a function graph of one variable $w^3$. It's very easy and I don't see any problem for this. – Michael Rozenberg Jun 29 '16 at 06:42