10

I'm looking for a proof to the following theorem:

For any $x,y\in R$: $x<y \Rightarrow x^3<y^3$

I'm trying this approach:

Let $z = x^3 - y^3 = (x-y)(x^2+xy+y^2) = z_1 z_2$

where $z_1 = x-y$ and $z_2 = x^2+xy+y^2$, so I must show that $z<0$. I know that $x<y \iff x-y<0$ so I just have to show that $z_2>0$ for any $x,y \in R$

I find it easy to prove that $z_2 > 0$ in the cases when $0<x<y$ and when $x<y<0$, but I can't prove it when $x<0<y$.

Here's my draft:

$x<y \iff x^2>xy$ $(1)$

$y<x \iff y^2>xy$ $(2)$

$(1)+(2)$ makes $x^2+y^2>2xy \iff x^2+xy+y^2>3xy < 0$

Any hints?

Thank you in advance!

  • 2
    Maybe I am missing somthing but is not the function x^3 one to one and strctly monotonic increasing over the reals. – Pete Jul 17 '14 at 19:30
  • @Pete: I think that is exactly what the OP wants to be able to prove. – TonyK Jul 17 '14 at 19:41

4 Answers4

21

$$ x^2 + xy + y^2 = x^2 + xy + \frac {y^2}4 + \frac {3y^2}4 = \left(x + \frac y2 \right )^2 + \frac {3y^2}4 > 0 $$

Kaster
  • 9,722
8

You don't need all that high-powered algebra. (By that I mean the factorization and the cleverness Kaster helped you with.) All you need is that if $a<b$ and $c>0$, then $ac<bc$. It's more boring, but it doesn't require any cleverness.

With that in mind, here's another proof, in three cases.

Case one: $xy>0$

If we multiply $x<y$ by the positive quantities $x^2$, $xy$, and $y^2$, we get the following three inequalities, respectively:

$$x^3<x^2y$$ $$x^2y<xy^2$$ $$xy^2<y^3$$

Putting these together gives the claim.

Case two: $xy<0$

In this case, we must have $x<0$ and $0<y$. Multiplying the first inequality by the positive quantity $x^2$ gives $$x^3<0$$ and similarly, multiplying the second inequality by $y^2$ gives $$0<y^3$$ Putting these together gives the claim.

Case three: $xy=0$

Finally, we have either $x=0$ or $y=0$ (but not both!). I'll let you deal with this case.

2

If $x < y$ then we can write $y = x + \epsilon, \epsilon > 0$. Now $$ \begin{align} y^3 &= (x + \epsilon)^3\\ &= x^3 + 3x^2\epsilon + 3x\epsilon^2 + \epsilon^3 \end{align} $$ So we just need to show that for $ \epsilon > 0$, $3x^2 + 3x\epsilon + \epsilon^2 >0$. The discriminant of this equation is $9\epsilon^2 - 12\epsilon^2 < 0$, so there are no real roots as the equation is positive all of the time, which proves that given $y > x, y^3 > x^3$.

Avraham
  • 3,237
  • this is nice too, but just to be totally explicit, this argument relies on the continuity of quadratic functions and the intermediate value theorem. (you need the IVT to go from the non-existence of roots to the fact that the function is always positive.) – symplectomorphic Jul 17 '14 at 21:19
2

Here's another way to prove that $x^2+xy+y^2>0$

$$x^2+xy+y^2>0\Leftrightarrow (x+y)^2>xy$$

If you think a little about that inequality you'll see it is always true given that $x \neq y$. To give a formal proof some case working is needed but it's bearable.

  • If $x$ or $y$ is zero then the inequality holds (RHS is positive, LHS is zero).
  • If $x$ and $y$ have opposite signs then the inequality also holds (RHS non-negative, LHS negative).

Now we prove the inequality holds if $x$ and $y$ have the same sign.

  • If both are positive, $|y|>|x|$:

$$(x+y)^2=(|x|+|y|)^2>|y|^2>|x|\cdot|y|=xy$$

  • If both are negative, $|x|>|y|$: $$(x+y)^2=(|x|+|y|)^2>|x|^2>|x|\cdot|y|=xy$$

It is not as elegant as Kaster's solution, but it's a systematic way of getting the job done.