0

Find the limit

$$\displaystyle\lim_{(x,y)\to(0,0)}\frac{-16x^3y^3}{(x^4+2y^2)^2}$$


Let $x=r\sin \theta$ and $y=r\cos \theta$.

$$ \begin{align} \lim_{(x,y)\to(0,0)}\frac{-16x^3y^3}{(x^4+2y^2)^2}&=\lim_{r\to0}\frac{-16r^3\cos ^3 \theta r^3\sin ^3\theta}{(r^4\cos^4\theta+2r^2\sin^2\theta)^2} \\ &=\lim_{r\to0}\frac{-16r^3\cos ^3 \theta r^3\sin ^3\theta}{r^4(r^2\cos^4\theta+2\sin^2\theta)^2}\\ &=\lim_{r\to0}\frac{-16r^2\cos ^3 \theta \sin ^3\theta}{(r^2\cos^4\theta+2\sin^2\theta)^2}\\ &=\frac{-16(0)^2\cos ^3 \theta \sin ^3\theta}{((0)^2\cos^4\theta+2\sin^2\theta)^2}\\ &=\frac{0}{(0+2\sin^2\theta)^2}\\ &=0 \end{align} $$

Does this logic follow? Can I ignore the case where $2\sin^2\theta=0$ meaning it is of the form $\frac{0}{0}$? How else would I approach this limit.

  • 1
    When a situation like this pops up, you're not allowed ignore the $\sin^2\theta$ left in the denominator because the limit could still exist or not exist. Usually for simple limits (but not always) the "worst case path" (the one most likely to create a distinct limit) when you have even powers in the denominator only is the one that homogenizes the denominator terms. In this case that would be the path $y=x^2$, and the limit still equals $0$ along this path - so it's worth trying to spend time looking for a proper squeeze theorem argument. – Ninad Munshi Mar 17 '21 at 09:21
  • Yes I thought that might be the case, just hoped their may be an easier solution. – Magna Wise Mar 17 '21 at 09:24
  • Hint : $|x|\leq (x^4+2y^2)^{1/4}$ and $|y|\leq (x^4+2y^2)^{1/2}$ . – Kelenner Mar 17 '21 at 09:30
  • You have not gotten the solution, but only proved you can't find the solution with that method because you got an intermidate $\frac{0}{0}$ – Aderinsola Joshua Mar 17 '21 at 10:19

4 Answers4

1

By AM-GM inequality we have that

$$x^4+y^2+y^2 \geq 3 x^{\frac{4}{3}}y^{\frac{4}{3}}$$

which we can use to say

$$ \frac{16|x^3y^3|}{(x^4+2y^2)^2}\leq \frac{16|x^3y^3|}{9x^{\frac{8}{3}}y^{\frac{8}{3}}} \leq 2\sqrt[3]{|xy|} $$

Thus the limit exists and equals $0$ by squeeze theorem.

Ninad Munshi
  • 34,407
  • If $xy=0$ you have a problem... – Kelenner Mar 17 '21 at 09:41
  • @Kelenner why? Everything looks good to go – Ninad Munshi Mar 17 '21 at 10:09
  • Have a look at your second inequality,, namely at $\frac{16|x^3y^3|}{9x^{\frac{8}{3}}y^{\frac{8}{3}}}$ – Kelenner Mar 17 '21 at 10:22
  • @Kelenner that doesn't answer my question. Everything is well defined, even if you want to handle $xy=0$ as a separate case. But I can assure you this is a well known technique used all the time. Have you seen holes in functions before? – Ninad Munshi Mar 17 '21 at 10:37
  • If $xy=0$, you have written $0/0$ ; in addition, if $xy=0$, you start from $x⁴+2y²\geq 0$, you cannot get anything from that... – Kelenner Mar 17 '21 at 10:39
  • @Kelenner this feels like you're stuck on your own crusade about a minor point and you didn't even read the rest of my comment that addressed it, but please feel free to tell me if my explanation was not adequate. Again, have you or have you not seen holes in functions before? – Ninad Munshi Mar 17 '21 at 10:49
  • I do not think that dividing by $0$ is a minor point... Sorry, I must go. – Kelenner Mar 17 '21 at 10:51
  • I can only agree that it seems foolish to omit the $xy=0$ case. I can't think of why this approach works. – Magna Wise Mar 17 '21 at 11:19
  • @OliverMurfett take $xy=0$ as separate case if you want, the function is $0$ on that set anyway since it seems like I have to state that. But I want to reiterate this is valid for all of $\Bbb{R}^2$ and I have not typed any new information. Again, have you seen holes in functions before or not? If the hole terminology is not familiar then rates of convergence might be. The $"0"$ in the numerator is faster than the $"0"$ in the denominator. – Ninad Munshi Mar 17 '21 at 11:21
0

Another approach is to homogenize degrees with $y=ux^2$ ($u$ variable).

After substitution $x^8$ can be simplified and we get $f(x,y)=\underbrace{\dfrac{-16u^3}{(1+2u^2)^2}}_{g(u)}\times x$

It is not difficult to see that $g$ is bounded, indeed $1+2u^2\ge 1>0$ so the denominator never annulates so $g$ is continuous on $\mathbb R$.

Also $|g(u)|\sim \frac{16|u|^3}{4|u|^4}=\frac 4{|u|}\to 0$ at infinity, therefore $g$ is bounded by some $M$.

We conclude that $|f(x,y)|\le |g(u)||x|\le M|x|\to 0$

zwim
  • 28,563
0

You already got the answer, in finding limits just avoid the intermidate $$\displaystyle\lim_{(x,y)\to(0,0)}\frac{-16x^3y^3}{(x^4+2y^2)^2}$$ You let $x=r\sin \theta$ and $y=r\cos \theta$.

$$ \lim_{(x,y)\to(0,0)}\frac{-16x^3y^3}{(x^4+2y^2)^2} = \lim_{r\to 0}\frac{-16r^2\cos ^3 \theta \sin ^3\theta}{(r^2\cos^4\theta+2\sin^2\theta)^2} = \frac{-16(0)^2\cos ^3 \theta \sin ^3\theta}{((0)^2\cos^4\theta+2\sin^2\theta)^2} = \frac{0}{(0+2\sin^2\theta)^2}$$

Remember that $\theta $ is also a variable $x=r\sin \theta$, so you can ignore the case when $2\sin^2 \theta$ is $0$, If fact we can deceive the equation by saying $\lim_{(\theta ,r) \to (0,0)}\frac{-16r^3\cos ^3 \theta r^3\sin ^3\theta}{(r^4\cos^4\theta+2r^2\sin^2\theta)^2}$, since $x$ and $y$ are approaching same objective $$\lim_{(x,y)\to(0,0)}\frac{-16x^3y^3}{(x^4+2y^2)^2} = \lim_{x \to 0} \lim_{y \to x}\frac{-16x^3y^3}{(x^4+2y^2)^2}$$

$$\lim_{(x,y)\to(0,0)}\frac{-16x^3y^3}{(x^4+2y^2)^2} = \lim_{x \to 0} \frac{-16x^3x^3}{(x^4+2x^2)^2} = \lim_{x \to 0} \frac{-16x^6}{(x^4+2x^2)^2}$$ As $x$ gets smaller the numerator shrinkes rapidly compared to the denominator, when you check L'hospitals rule after repeated sections the numerator would vanish meaning the limit is $0$

  • What you say about $\theta$ is an excellent point, however you came to the wrong conclusion. $\theta$ can be any real number at $(0,0)$, so saying the limit must exist if you take $\theta$ to approach $0$ is incorrect. The takeway from $\theta$ being a variable in its own right is that the limit as $r\to0^+$ must exist uniformly (i.e. independently of $\theta$, but a more precise way of phrasing that). You plugged in the path $y=x$ which is good, but remember one path going to a value doesn't mean all the paths have to go to that value. – Ninad Munshi Mar 17 '21 at 11:15
  • This feels alot like taking the limit along a certain path (y=x in this case), which does not prove the limit exists. Is this method mathematically rigourous? – Magna Wise Mar 17 '21 at 11:16
0

These things get confusing, I like to work out the extreme values using Lagrange multipliers. Here, I demand $x > 0, y > 0,$ maximum value of $$ \frac{x^3 y^3}{(x^4+2y^2)^2} $$ when the denominator is thought of as the constraint. Go through the gradients being parallel, the largest value occurs when $$ 24 x^2 y^2 (x^4 + 2 y^2)(x^4 - y^2) = 0 \; . \; $$ The letters are real and positive, so $y=x^2$ gives the maximum for fixed denominator. This is then $$ \frac{x^9 }{(x^4+2x^4)^2} = \frac{x^9 }{9x^8} = \frac{x }{9} $$ That is the biggest (my) fraction can be, in the first quadrant.

I was disappointed to find that the apparent inequality suggested does not hold in general;safer to say that, when the denominator is $C>0,$ we know $x < C^{1/8}$ in the first quadrant. Then, let's see, the ratio is less than $$\frac{ C^{1/8}}{9}$$

Will Jagy
  • 139,541
  • This is possibly the best answer I've seen yet, I might use the $\frac{x}{9}$ upper bound and use the sandwich theorem. (And obviously, prove the inequality). – Magna Wise Mar 17 '21 at 20:58