1

Let $x_1,x_2,\ldots,x_n$ be arbitrary real numbers. Prove the inequality $$\frac{x_1}{1+x_1^2} + \frac{x_2}{1+x_1^2 + x_2^2} + \cdots + \frac{x_n}{1 + x_1^2 + \cdots + x_n^2} < \sqrt{n}$$


I found a solution here which says:

$$\frac{x_1}{1+x_1^2} + \frac{x_2}{1+x_1^2 + x_2^2} + \cdots +\frac{x_n}{1 + x_1^2 + \cdots + x_n^2} < \sqrt{n}$$ $$ \iff \left(\frac{x_1}{1+x_1^2} + \frac{x_2}{1+x_1^2 + x_2^2} + \cdots +\frac{x_n}{1 + x_1^2 + \cdots + x_n^2}\right) ^2< n$$ but we have: $$ \left(\frac{x_1}{1+x_1^2} + \frac{x_2}{1+x_1^2 + x_2^2} + \cdots +\frac{x_n}{1 + x_1^2 + \cdots + x_n^2}\right)^2 \le n\left(\frac{x^2_1}{(1+x_1^2)^2} + \frac{x^2_2}{(1+x_1^2 + x_2^2)^2} + \cdots +\frac{x^2_n}{(1 + x_1^2 + \cdots + x_n^2)^2}\right)$$ $$ \le n\left(\frac{x^2_1}{1\cdot (1+x_1^2)} + \frac{x^2_2}{ (1+x_1^2)(1+x_1^2 + x_2^2)} + \cdots +\frac{x^2_n}{(1 + x_1^2 + \cdots + x_{n-1}^2)(1 + x_1^2 + \cdots + x_n^2)}\right)=n\left(1-\frac{1}{1 + x_1^2 + \cdots + x_n^2}\right)<n$$


My query is that they claimed $$ \frac{x_1}{1+x_1^2} + \frac{x_2}{1+x_1^2 + x_2^2} + \cdots +\frac{x_n}{1 + x_1^2 + \cdots + x_n^2} < \sqrt{n}$$ $$\iff \left(\frac{x_1}{1+x_1^2} + \frac{x_2}{1+x_1^2 + x_2^2} + \cdots +\frac{x_n}{1 + x_1^2 + \cdots + x_n^2}\right) ^2< n$$ but the LHS of the first inequality need not to be positive. We know $x^2$ is not strictly increasing in all reals, then how did they conclude the second line from the first one?

  • You only need $\Leftarrow$ anyway instead of $\iff$ to prove your inequality. – Aig Feb 20 '24 at 14:10
  • 1
    If the LHS is negative the inequality holds anyway, so there is only something to prove if the LHS is positive. If we can prove the squared inequality for the positive case then we are done. It looks like they proved that the squared inequality is true in all cases, so they actually proved the stronger statement $|LHS|<\sqrt{n}$, or equivalently $-\sqrt{n}<LHS<\sqrt{n}$. I agree that the $\iff$ symbol is incorrectly used in their solution. – Jaap Scherphuis Feb 20 '24 at 14:40
  • FYI, it should be zhaobin@AoPS's solution. See here. – River Li Feb 21 '24 at 00:19

0 Answers0