6

How to find the asymptotes of the implicit function $$8x^3+y^3-6xy-3=0?$$

user64494
  • 5,811
  • 1
    What methods can you use? There is a change of variable to make the xy-term disappear. Then you can graph and go back to your change of variable. – DBFdalwayse Aug 24 '13 at 07:27
  • @DBF: Could you realize it? – user64494 Aug 24 '13 at 07:32
  • This is what I'm trying: to rotate the axes: $x'=xcos\theta-ysin\theta ; y'=xsin\theta +ycos\theta$ , expanding and setting the mixed terms to zero. I'm working on it :http://www.wolframalpha.com/input/?i=8%28cosa-ysina%29%5E3%2B%28xsina%2Bycosa%29%5E3-6%28xcosa-ysina%29*%28xsina%2Bycosa%29%3D – DBFdalwayse Aug 24 '13 at 07:44
  • @DBF8: Unfortunately, I see nothing there. – user64494 Aug 24 '13 at 07:46
  • Sorry; hopefully this is better: http://www.wolframalpha.com/ : you can enter the terms with the change of variables and find the angle $\theta$ that would make the mixed $x^ny^m$ terms disappear. – DBFdalwayse Aug 24 '13 at 07:52
  • 1
    I will be waiting for a serious and constructive answer. – user64494 Aug 24 '13 at 07:55
  • I'll try my best and will get back to you. – DBFdalwayse Aug 24 '13 at 07:59
  • @DBF nice one :) – Kaster Aug 24 '13 at 08:35
  • Thanks, Kaster, but I ended up bogged down. User64494: I have been stuck for long. The best I can offer you is to go to Wolfram and ask to plot the curve. There clearly is no rotation that will make the xy-term disappear, since this curve is (very clearly) not a conic. Sorry; let me withdraw; will try to come up with something else. – DBFdalwayse Aug 24 '13 at 09:06

3 Answers3

6

The article you need to thoroughly put your eyes on is this Asymptote: Algebraic curves. So, let's do it step by step

1) Split your polynomial into homogeneous ones \begin{align} P_3(x,y) &= 8x^3+y^3 \\ P_2(x,y) &= -6xy \\ P_0 &= -3 \end{align} $$ f(x,y) = 8x^3+y^3-6xy-3 = P_3(x,y) + P_2(x,y) + P_0 = 0 $$ 2) Try to decompose $P_3(x,y) = Q(x,y)$ to the form of $(ax-by)Q_2(x,y)$, where $Q_2$ is another homogeneous polynomial. In this particular case it's quite easy to do, since $$ P_3(x,y) = 8x^3+y^3 = (2x+y)\left( 4x^2-2xy+y^2\right ) $$ so $a = 2$ and $b = -1$.

3) Find the values of $Q_x'(b,a)$ and $Q_y'(b,a)$ \begin{align} Q_x'(b,a) &= \left . 24x^2 \right |_{x = -1} = 24 \\ Q_y'(b,a) &= \left . 3y^2 \right |_{y = 2} = 12 \end{align} so, they don't vanish simultaneously, and therefore $$ Q_x'(b,a) x + Q_y'(b,a) y + P_2(b,a) = 24 x + 12 y + 12 = 0 $$ is an asymptote. Latter can be simplified as $$ y = -2x - 1 $$ To make sure of it, you can draw that expression's plot.

asym

PS: There are actually several plots, as a demonstration that asymptote doesn't depend on $P_0 = 3$, it can be $0$ or $100$.

Kaster
  • 9,722
6

The asymptotes of an algebraic curve are simply the lines that are tangent to the curve at infinity, so let's go through that calculation.

First, we find where your curve meets the line at infinity. We homogenize to $(X:Y:Z)$ coordinates, so that $(x,y) = (X:Y:1)$. The equation is now

$$8X^3+Y^3−6XYZ−3Z^3=0$$

and the line at infinity is the line $Z=0$. Plugging that in and solving over the reals

$$ 8X^3 + Y^3 = 0 \\ Y = -2X$$

so $(1:-2:0)$ is the only (real) projective point where the curve meets infinity. (There will be two complex asymptotes as well, but we can't see them if we're only looking at geometry with real coordinates)

To find the tangent line, we note that the differential of the tangent line is the same as the differential the curve (more or less because the tangent lines have to point in the same direction). So we take the derivative:

$$ (24X^2 - 6YZ) dX + (3Y^2 - 6XZ) dY - (6XY + 9Z^2) dZ $$

and evaluate at $(1:-2:0)$ to get

$$ 24 dX + 12 dY + 12 dZ$$

and so the tangent line should be given by the equation

$$ 24 X + 12 Y + 12 Z = 0 $$

or, dehomogenizing:

$$ 24x + 12y + 12 = 0 \\ y = -1 - 2x $$

This is, of course, the same calculation as in the other answer, but if you're familiar with projective geometry (or are willing to self-study how it works), it explains how the calculation works.

2

I have seen you are interested in doing problems by Maple so, the following codes may help you machineary:

[> f:=8*x^3+y^3-6*x*y-3:
   t := solve(f = 0, y):
   m := floor(limit(t[1]/x, x = -infinity));

$$\color{blue}{m=-2}$$

[> h:=floor(limit(t[1]-m*x, x = -infinity));

$$\color{blue}{h=-1}$$

Mikasa
  • 67,374
  • I think that approach works with a broader classes of functions than polynomials. I find $$m:=limit(t[1]/x,x=-infinity): simplify(m);$$ more transparent. Could you explain how to insert a code in SE? – user64494 Sep 16 '13 at 11:24
  • $\quad \ddot \smile \quad$ +1 – amWhy Sep 16 '13 at 12:14
  • I still have a couple of recent-ish answers with big fat $0$ $\uparrow$'s...one's even accepted...Oh well, maybe will delete a couple of – amWhy Sep 16 '13 at 13:10