2

Let real $a,b,c,d > 0$. Show that \begin{align} \frac{1}{64}(a^{15} + b^{15} + c^{15}+ d^{15})^4 &\ge \sum_{\rm {cyc}} (a b^2 c^4 d^8 )^4 \\ &= (a b^2 c^4 d^8 )^4 +(b c^2 d^4 a^8 )^4 +(c d^2 a^4 b^8 )^4 +(d a^2 b^4 c^8 )^4 \end{align} This is obviously homogeneous. Equality appears at $a=b=c=d$, I haven't found other equality points. Applying AM-GM inequality to the LHS is too crude because it doesn't take into account the rising exponentials $(1,2,4,8)$ on the RHS. Indeed, AM-GM leaves to prove \begin{align} 4 &\ge \sum_{\rm {cyc}} \frac{c\cdot d^{17}}{b^7 \cdot a^{11}} \end{align} which can be made to fail easily, e.g. by setting $a=b=c =0.01 \cdot d$.

Andreas
  • 15,175
  • If you know that your problem is equivalent to another (known but unsolved) problem (as I understand it from your below comment) – why don't you make that clear in the question? – Martin R Jan 30 '20 at 07:48
  • @MartinR Making the equivalence clear usually has the advantage of linking to other known approaches. Since there are no satisfying approaches available for the last five years, I reckon that the link is rather unsupportive or even demotivating. Instead, a new form of the problem may encourage people to go about the task with fresh, uninfluenced ideas. – Andreas Jan 30 '20 at 09:06

2 Answers2

1

Let $a b^2 c^4 d^8=t,$ $b c^2 d^4 a^8=z,$ $c d^2 a^4 b^8=y$ and $d a^2 b^4 c^8=x$.

Thus, $$\sum_{cyc}\frac{x^2}{y}=\sum_{cyc}\frac{d^2a^4b^8c^{16}}{cd^2a^4b^8}=\sum_{cyc}c^{15}$$ and we need to prove that $$\frac{x^2}{y}+\frac{y^2}{z}+\frac{z^2}{t}+\frac{t^2}{x}\geq\sqrt[4]{64(x^4+y^4+z^4+t^4)},$$ which is a known problem.

For example, see here: How to prove $\frac{1}{4}(\frac{a^2}{b}+\frac{b^2}{c}+\frac{c^2}{d}+\frac{d^2}{a})\ge \sqrt[4]{\frac{a^4+b^4+c^4+d^4}{4}}$

  • Thank you Michael. However, this known problem is known already for 5 years without a sufficient answer, and also new approaches (see yesterday's https://math.stackexchange.com/questions/3525260/) haven't produced answers yet. So I was trying to give it a new direction, instead of referring to the old questions. – Andreas Jan 29 '20 at 20:37
0

The answer to the corresponding problem (see Michael's comment) has been provided by me at math.stackexchange.com/a/3566860/317854

Hence the problem is solved.

Andreas
  • 15,175