2
$\begingroup$

This is exercise 18. d) from Spivak Calculus. Some things I have proven from previous problems that I was using to help me figure this out are:

$x^2+y^2\geq 2xy$; equality when x and y are both 0

$x^2+y^2\geq0$

I believe I know the answer however I don't think my method is leading to a valid proof.

Here is my process, I rearranged:

$x^2+axy+y^2>0$

$\Rightarrow x^2+y^2>-axy$

$\Rightarrow (x^2+y^2)/xy>-a$

I then considered what value -a, must be in order for this inequality to not hold:

So I want to find where $(x^2+y^2)/xy<-a$

Suppose $xy\geq0$

$x^2+y^2\geq2xy$ $\Rightarrow (x^2+y^2)/xy\geq2$

Thus $-a>(x^2+y^2)/xy\geq2$

Therefore if $(x^2+y^2)/xy<-a$ then $-a>2$. It follows that if xy<0 that -a<-2.

This is about where I'm stuck. I know I can't simply say that since $(x^2+y^2)/xy>-a$ is not true when $-a>2$ or $-a<-2$. That I then know the interval where it is true is $-2<-a<2$.

There's some other odd things I thought about but I'm not sure if they go anywhere. I've spent some time thinking about how x^2+y^2 changes compared to $axy$, how many counterexamples there are when a>2. I also had proven $x^2+xy+y^2>0$ as the sum of 2 squares previously but I don't think it helps me here.

$\endgroup$
  • 1
    $\begingroup$ Try completing the square on $x^2+axy+y^2$ $\endgroup$ – sharding4 Aug 5 '17 at 20:48
  • 1
    $\begingroup$ @sharding4 Wow, was much simpler then I was making it out to be. $\endgroup$ – AColoredReptile Aug 5 '17 at 22:28
1
$\begingroup$

Suppose $y\ne 0$. We want that $$\forall x\in \Bbb R \;\; f (x)=x^2+axy+y^2>0.$$

with $f (0)=y^2>0$.

the discriminant must be $<0$.

$$\Delta=y^2 (a^2-4)<0$$ which gives

$$-2 <a <2$$

$\endgroup$
  • 1
    $\begingroup$ Problem with your polynomial? OP has $x^2+axy+y^2$ $\endgroup$ – sharding4 Aug 5 '17 at 20:54
  • $\begingroup$ Except the typo with $a\mapsto 2a$, this is much cleaner. Well done (I took the liberty to fix it ) $\endgroup$ – Paolo Leonetti Aug 5 '17 at 20:56
  • $\begingroup$ @PaoloLeonetti Thanks . feel free. $\endgroup$ – hamam_Abdallah Aug 5 '17 at 21:01
1
$\begingroup$

Assume $(x,y) \neq (0,0)$ and consider that $$ x^2+axy+y^2=\frac{(2x+ay)^2+(4-a^2)y^2}{4}. $$ This is always $>0$ if: (i) $(4-a^2) \ge 0$. (ii) $(2x+ay)$ and $y$ are not both $0$.

Now (i) means $|a| \le 2$. At this point: if $|a|=2$ then you have $(x\pm y)^2>0$ which is false in general. Otherwise $|a|<2$: here, $2x+ay=y=0$ if and only if $x=y=0$, which has been excluded by hypothesis.

Hence, the answer is $a \in (-2,2)$.

$\endgroup$

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.