How to prove $\displaystyle \frac{x^2}{y+z}+\frac{y^2}{x+z}+\frac{z^2}{x+y}\leqslant\frac{2(x^2+y^2+z^2)}{x+y+z}$ where $x,y,z$ are all positive real numbers? The hint was to use the Schwarz inequality.

  • 2
    $\begingroup$ Maybe you could try the Schwarz inequality instead? ;-) $\endgroup$ – WimC Mar 31 '13 at 15:45
  • $\begingroup$ Lots of inequality techniques are in Lohwater's unpublished "Introduction to Inequalities" (the link might be bad, look for it in Google if it doesn't work). $\endgroup$ – vonbrand Mar 31 '13 at 15:50
  • 4
    $\begingroup$ Please check the Q. Try $x=2, y=1, z=0$ for e.g. $\endgroup$ – Macavity Mar 31 '13 at 15:59
  • 2
    $\begingroup$ Let $x+y+z = 3$. When $x \to 3^-$, $y,z \to 0^+$, it's clear that LHS is unbounded while RHS is bounded. $\endgroup$ – user27126 Mar 31 '13 at 16:32
  • $\begingroup$ You must have mistyped the expressions in the inequality because with $x=2, y=1,z=0.1$ the left hand side is greater than the right hand side as Macavity mentioned. On the other hand with $x=y=z=1$ the left hand side is less than the right hand side. $\endgroup$ – user782220 Apr 10 '13 at 1:22

The reverse inequality holds, and can be shown in the following way:$$~$$ Multiply both sides of the original inequality with an $(x+y+z),$ and use $\dfrac{x^2(x+y+z)}{y+z}=\dfrac{x^3}{y+z}+x^2,$ so that it suffices to check that $$\frac{x^3}{y+z}+\frac{y^3}{z+x}+\frac{z^3}{x+y}\geq x^2+y^2+z^2,$$ Which is a direct application of the Cauchy-Schwarz inequality in accordance with $(xy+yz+zx)\leq(x^2+y^2+z^2)$: $$\sum_{cyc}\frac{x^3}{y+z}\sum_{cyc}x(y+z)\geq (x^2+y^2+z^2)^2,$$ And we are done. Equality holds in the original inequality if and only if $x=y=z,$ or $x=y,z=0$ and its cyclic permutations. $\Box$

  • $\begingroup$ What were the two vectors you used in your application of the Cauchy-Schwarz inequality? $\endgroup$ – user782220 Apr 9 '13 at 10:13
  • $\begingroup$ Somewhere in the implied algebraic manipulations there is a flaw because the inequality is not true in both directions. $\endgroup$ – user782220 Apr 10 '13 at 1:19
  • $\begingroup$ The two sets of terms I used are $\sqrt{\frac{x^3}{y+z}}$ and $\sqrt{x(y+z)}.$ Maybe that form of C-S is better known as "Titu's lemma," since $\frac{x^3}{y+z}=\frac{(x^2)^2}{x(y+z)}$ etc. I think the manipulations are also correct. Please rectify me if I am wrong. $\endgroup$ – Potla May 7 '13 at 12:57

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.