1
$\begingroup$

let $x,y,z>0$ and such $$x+y+z=1$$ Find minimum of the $$\dfrac{xy}{x^5+xy+y^5}+\dfrac{yz}{y^5+yz+z^5}+\dfrac{xz}{x^5+xz+z^5}$$

I have find this maximum

note $$\dfrac{xy}{x^5+xy+y^5}\le -243\dfrac{x+y}{841}+\dfrac{23031}{1682}$$ see http://www.wolframalpha.com/input/?i=xy%2F%28x%5E5%2Bxy%2By%5E5%29%2B243*%28x%2By%29%2F841-23031%2F1682

But for minimum value,I can't,(maybe use AM-GM) Thank you,

$\endgroup$
  • $\begingroup$ I can't see it being anything except $81/29.$ If all three have to be positive, then there will be trade-offs if $x \neq y \neq z$ because of the cyclic symmetry of the function to minimize. So $x = y = z = 1/3.$ But I can't prove it. $\endgroup$ – John Jan 8 '14 at 4:14
  • $\begingroup$ For $g(x,y)=xy/(x^5+xy+y^5)$ if $\partial_x^2 g(x,y) < 0$ for $0<x,y\leq 1$, which some graphing may suggest, then I think there is an argument using Lagrange multipliers that $x=y=z$ is where the min occurs. But it is too late now... $\endgroup$ – abnry Jan 8 '14 at 4:41
  • $\begingroup$ Lagrange multipliers looks hairy. I'm guessing there's a trick. $\endgroup$ – John Jan 8 '14 at 5:09
  • $\begingroup$ @John Lagrange multipliers work if $g_x(x,y)$ is one-to-one for all $y<1$. All the symmetries work out otherwise in just a fine manner (with some argument). $\endgroup$ – abnry Jan 8 '14 at 5:48
1
$\begingroup$

The max is $\dfrac{81}{29}$ and min is $\dfrac{.25}{2*(0.5)^5+.25}$, a simple method is let two varies equal$(x=y)$ and have a function of $x$, see graphic below:

enter image description here

edit :to prove max, consider $xy \le \dfrac{x^2+y^2}{2}, x^5+y^5 \ge 2(\dfrac{x^2+y^2}{2})^{\frac{5}{2}},\dfrac{xy}{x^5+xy+y^5}\le \dfrac{x^2+y^2}{x^2+y^2+4(\dfrac{x^2+y^2}{2})^{\frac{5}{2}}}=\dfrac{p}{p+4(\dfrac{p}{2})^{\frac{5}{2}}}=f(p)$

$f(p)$ is concave and mono decreasing function and $x^2+y^2+z^2 \ge \dfrac{1}{3}$ ,so we can quickly know the max will be got when $x=y=z$

$\endgroup$
  • $\begingroup$ Hello,the max is how prove it? becasue I think $$\dfrac{xy}{x^5+xy+y^5}\le\dfrac{23031}{1682}-\dfrac{243}{841}(x+y)$$ is not easy prove $\endgroup$ – math110 Jan 8 '14 at 7:42
  • $\begingroup$ @math110 this value is wrong!$\dfrac{xy}{x^5+xy+y^5}< 1 \implies $ max $<3$,but $\dfrac{23031}{1682}>13,x+y<1 \to $ Max $>12$ $\endgroup$ – chenbai Jan 8 '14 at 7:51

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.